Documentation: obs_common_subset typos, corrections

[kdraeder]

What's the issue?

📚 obs_common_subset.html has several minor mistakes I'd like to note as I'm reading.

Where did you find the issue?

assimilation_code/programs/obs_common_subset/obs_common_subset.html
https://docs.dart.ucar.edu/en/latest/assimilation_code/programs/obs_common_subset/obs_common_subset.html#program-obs-common-subset

What needs to be fixed?

"obs_selection_tool" is named "obs_selection", and the text is not a link.

"list of lists of filenames" would be clearer as "list of files, each containing a list of obs_seq files"

ls exp3/*/obs_seq.final > exp3list
cat exp2list

filename_seq_list = 'exp1list', 'exp2list', exp3list' Needs a ' before exp3list

In the table of namelist entries we might want to change "printed every Nth set of obs"
to "printed at multiples of print_every"

Building: "Most $DART/models/*/work directories" would more accurately (but less robustly) be
"Most $DART/models/*/work/quick_build.sh"

Anything else we should know?

[kdraeder]

Some of the documentation is a bit misleading.
In the obs_common_subset.f90 header is written:

!>  The obs to be assimilated or evaluated can be selected by namelist
!>  at run time if they are different, and different experiment settings
!>  can be selected by other namelist values.

The ensemble size cannot differ between the files because it affects the structure of the files,
not just the contents (numbers, character strings, ...). There's a check of ensemble sizes,
which stops the program if they differ. It would be helpful to change the error message from

      write(msgstring2,*)'Different numbers of data copies found: ', &
                         num_copies1, ' vs ', num_copiesN

to

     write(msgstring2,*)'Different numbers of data copies found: ', &
                         num_copies1, ' vs ', num_copiesN, &
                         '.  Remove all ensemble copies from all files using obs_sequence_tool.'

or put the "Remove" line into msgstring1 and add msgstring1 to the error_handler call.

---

And this feature raises raises a question;

!>  If the next input obs don't match exactly (identical types, times, locations,
!>  and DART QC) they are discarded.

If the obs_seq files have the same obs, but in different orders, then some common obs won't be found
and the resulting obs_seq.common files will be misleading. It seems that the program should stop,
with instructions for how to fix the problem (if it's fixable).