CDMS: fix MOLWT column starting position

[kmaitreys]

Now the molecule tag 100501 and others work. Earlier they were crashing with

ValueError: invalid literal for int() with base 10: '13-'

Because that - is part of the signed TAG.

[bsipocz]

RTD failure is unrelated and will be fixes outside this PR.

[keflavich]

We'll need some tests for this error. I can confirm that 100501 does not work right now:

CDMS.query_lines(10*u.GHz, 100*u.GHz, molecule='100501', parse_name_locally=False)

I tested one other case, and it worked both before & after:

CDMS.query_lines(10*u.GHz, 100*u.GHz, molecule='HCCCN', parse_name_locally=True)

so I'm going to approve this, but on my to-do list is to systematically test all molecules for failures like this.

[bsipocz]

In addition to putting it on the to-do, please do open an issue for it here, too.

[bsipocz]

Test failures are related

[kmaitreys]

Yeah, I'm sorry, the last commit I pushed actually broke a bunch of other molecules. So far from my testing/debugging, it looks like just changing the column numbers will not help, because the format itself is not consistent for all molecules. It needs a bit larger refactor in some parts.

I'm still up for fixing this, but couldn't due to my own work. If it's possible, I would like this to be kept open for a few weeks more.