Fix wrong crystalform-site tag generation

[phraenquex]

@kaliif points out in #1322 (this comment) that the CrystalForm site is generated wrong, because the original spec #1277 was wrong.

(It was in fact sensationally wrong, no idea how @kaliif magically made sense of what @phraenquex put there.)

This is what the spec should be:

• Short name: F1a
• Long string: A71EV2A-x0450/A/201 - small cell <xtalform_sites.xtalform_site_id>
• F1 - same as crystalform.id
• a - number a-to-z
  • sort crystalform-sites by the the total number of observations in the corresponding canonical site
  • for the top-ranked crystalform-site, use lowercase of chain id (the "A" from the example long string above)
  • for the next-ranked site, continue <a-z>, <aa-az,ab-az...>
  • etc. down the ranking.

(I think this is right - I'll only know if it's wrong once you've tried it.)

[kaliif]

@phraenquex the instruction in the linked comment must have been superseded because the way the tags are generated is pretty close to what you laid out here. The biggest change here would be the ordering of sites, at the moment they are labeled with letters in the order they come in, but above it says the letter should correlated with the number of observations in canon site.

This means that if new crystalform site objects are added between the uploads, old tags will be overwritten if their order changes. Is this the expected behaviour?

[phraenquex]

@kaliif no that's the spec for the first upload. For subsequent uploads, you just stick things on at the end. (It's a very VERY rare cornercase - mostly, everything relevant for the crystalform will have been nailed down for the first upload.)

But no, it doesn't sound like you've interpreted my spec correctly - can you please take another careful read, and check with @ConorFWild or @mwinokan if unsure. Specifically, the numbering for the top-ranked site is different than for the others.

[mwinokan]

@kaliif says it would help to have a concrete example. @mwinokan to talk to @ConorFWild

[kaliif]

I've merged what I could understand from the specs to staging, once this becomes available, you can try it out. If I got it wrong again, then yes, I am going to need a clear spec with examples.

[phraenquex]

Well done! I'm not pretending I was definitely comprehensible...

@mwinokan could you arrange for a fesh upload of Jasmin's CHIKV.

[phraenquex]

@mwinokan to alert @phraenquex when the next set of data have been uploaded.

[mwinokan]

@phraenquex please see the table here, there are multiple newly aligned targets with which to test