Report absorption features on output of 'save fine continuum' command

See merge request !37

Previously, the header produced by the command used the words 'Species'
and 'Opt Dep Spec'.  The former is outdated, given that we now report
the spectral line label (including wavelength), and the latter is not
clear to the user, given that we refer to the single-line optical depth
as Specific only in the source.

Here, the former is replaced with 'SpecLine' and the latter with
'Single Line Opt Depth'.

The previous implementation of the code block in LineStackCreate() that
quantifies line overlap used an arbitrary value of 20 as the maximum
number of lines that could overlap in a bin.

During review, Gary questioned the choice.  This ad hoc value was picked
as a high enough limit to carry out the task, given that currently up to
14 lines overlap in a bin.

The previous implementation was also inefficient in that it went over
rfield.fine_lstack map several times to produce the report.

Here, the overlap across rfield.fine_lstack is established ahead of
time, and each fine continuum index in the map is visited only once for
the report.  The ad hoc value of 20 is removed.

A (disabled) block of code was added at the end of LineStackCreate()
that worked its way through the map of fine continuum bins and line
stack indices to quantify the amount of overlap present.  The block
did not carry a comment.  During review, it was deemed necessary to
explain its intent.  This is done here.

Also, on the output of bins with overlap greater than 10 lines, the
line stack index is now reported to show that the repeating species
labels are indeed genuine.

The previous implementation reported the line center optical depth.
Here, we report the mean optical depth to keep the output the same as
the specific optical depths reported on the main output with the command
'print line optical depths'.

Update scripts and source for release preparation

See merge request !36

The previous implementation reported only a feature's species in
spectral notation, e.g., 'Fe17'.

This is useful if cells are not merged and the energy is reported in
wavelength units (say, Angstrom), because the exact wavelength is
reported in the first column.

Merging cells eliminates the wavelength information, and if the number
of cells merged together is high, the output can be confusing.

Report the entire spectral label, including the wavelength, to eliminate
ambiguity in the output.

The previous implementation reported only the most opaque line per bin.
Merging cells may eliminate important lines from the output.

All lines contributing opacity to a cell are now reported, if their
optical depth exceeds a limit.

The previous implementation of 'save fine continuum' accumulated the
first bin separately from any requested others (9 more, by default).
That forced code duplication relating to identifying the most opaque
line in the bin.

Restructure the code to treat all bins at once; eliminate duplication.

Anna Ogorzalek posted a query on the Cloudy group about identifying
absorption features in the output of the 'save xspec mtable' command.

A partial answer would be to identify features due to the lines alone.

The present adds two columns to the output of 'save fine continuum'.
The first identifies the species whose line is centered at that bin,
and the second reports the optical depth due that line alone (i.e., not
the total including any neighboring lines).

Protection against division by zero cooling.

See merge request !35

The demotes cautions that parts of the SED are zero to notes.
The following wc on auto shows that we have a large number of
these.  They are benign - they occur if we use a blackbody
that does go to zero far out in the Wien tail.  They are a
minor note about the SED rather than something deserving of
raising a caution.

the number of cautions in tsuite/auto went from

Alien20:auto:wc ~gary/cloudy/cautions.txt
  5267  72721 544951 /home/gary/cloudy/cautions.txt

to

Alien20:auto:wc cautions2.txt
  3846  64015 460146 cautions2.txt

The command in question was 'save chemisty rates'.

Missing references lead to undefined number of rows when generating the
TeX table.  Add guards against that possibility.

Ease use of the script by making sure it is not repeatedly asking for
the ADS token.  Store after first call, use stored in subsequent.

There was a bug in the previous implementation, which amounts to issuing
ADS BibTex queries in non-interactive mode.  Shut them down.

The previous ADS BibTeX service (cgi-bin/nph-bib_query) no longer
exists.

As a result, storing which of the three databases (AST, PHY, and PRE)
is the origin of the BibTeX data has become obsolete.  Remove all
instances of use of the databases, and relevant data fields.

Instead, the new API requires users to supply a token in order
to query the server from the command-line.  Update to current method.

Use of the new API requires that an ADS token be used.  Users need
supply one only in interactive runs.

This occurs when computing the number of lines in the TeX table.

The previous implementation reported a file as empty based on the
presence of a 'Reference' line.

Some files that contain the line but are not followed by any references,
were not reported as empty.  This is now fixed.

With the previous implementation, the data structure that is written out
as a JSON file is pruned of no longer existing references only in
interactive mode.

Now it is always done.  Fixed.

With the previous implementation NIST references were processed only in
interactive mode.  This prevented updating dates on NIST refs.

These are now updated, if needed, based on any available dates.

Some Stout files miss the 'Reference' line, and yet others contain data
on that line.  The previous implementation did not handle them
correctly.

Perl module Text::BibTeX already defines them.  Previous code caused
warnings such as:
	'warning: overriding existing definition of macro "dec"'
to be issued.  This is no longer the case.

The file contains several non-BibTeX entries as plain text, which cause
warnings to be emitted.  Replace some \bibitem entries with their formal
BibTeX record from ADS.  Comment the rest out with '%'.

These protect against division by zero.  Problems were exposed
by the lab sim posted in gtests / bugs / Mancini.  The
sim does get now much farther along but aborts when the line
stack becomes misaligned.

The sim is a lab plasma, so it is far from our home base, but we
should be able to compute it.  It is non-equilibrium.  It is
unclear why the cooling is zero - perhaps the total is net heating.

The changes seem safe since they are have never been needed and
will have no effect in most cases.