Fixed column's example width bar.

When creating or editing a column column, the example width bar didn't change anymore when you set the column's width.
- JQuery now needs val() instead of attr('value') to get and set the input's value.

Added support for clinical classification columns.

Added support for VariantOnGenome/ClinicalClassification and VariantOnGenome/ClinicalClassification/Method in the submission API.
- The method will be filled in, if we find something that equals one of the methods in the pathogenicity comments. There is currently no VarioML field for it.

Created the Variant Validator object and a variant verification script.

- When it remotely looks like an RNA description, keep the original one. Otherwise, overwrite.
- Also fixed a previous regular expression and applied a forgotten fix to the use of lovd_fixHGVS().

- There was a risk for an endless loop.

- Let it retry the call without the transcript, instead.

The variant fixing script now fills in empty VOT DNA, RNA and protein fields, also if just one of these is empty.

- Variants with ins[NC....] are now recognized and silently ignored.
- Also, connect to the VV test endpoint, the production endpoint doesn't support our features yet.

- At this time, the latter is the only endpoint that actually works with this VV class.

- In theory, the processing_error can be present, but I currently cannot reproduce that.

- Added a FIXME about its implementation in verifyGenomic(). We're still waiting for that, it throws an ESYNTAX there now.

- There's three different messages it can throw now. This new one is for 3' UTR variants too far from the transcript (c.*5000del).

- Do ignore WCORRECTED and WROLLFORWARD, they are truly harmless.

- The ^ symbol is used as a logical "or"; these descriptions are not supported.
- The [(number)] sequence, like [3], is used for repeats. These descriptions are not supported.

- If variants partially map outside of a transcript, VV currently does not return a transcript mapping, but also no liftover. As such, we can do nothing.
- See https://github.com/openvar/variantValidator/issues/173.
- Silently skipping these variants for now.

When checking the RNA field with lovd_fixHGVS() and lovd_getVariantInfo(), make sure you pass lovd_fixHGVS() the right prefix.

- They should also be allowed to pass intronic positions.

- Since we prevent the checking of very large variants, this shouldn't just happen anymore. I want to see it when it does.

- Also, unhandled errors how show more information.

- Throw a ESIZETOOLARGE, instead.
- Also, fixed some typos.

Fixed admin tests.

- assertContains() will, in a later PHPUnit version, stop supporting string comparisons.
- Minor change to the Download verification test, making the setUp() call a bit more logical.

- Now testing for the possibility of two strings matching.

- The original unit test, loaded in the "authorization_tests" suite, did not test for Colleague rights. This replaces their Submitter tests.

- Also, since we have no user ID #00007 (we use the Submitter #00006 account as a Colleague), remove all mentions of it.

- When I only have one session ID, don't try to create a new one.

- The cookies were a mix of old and new cookies, and we can't assume the first one matches our current LOVD.
- Compare cookie array before and after logging in, to determine the correct current LOVD's ID.

- Assigning user 6 (Submitter) as a Colleague of user 5 (Owner).

- Fixed the download file for the download verification to match.
- Mutalyzer's test instance seems to know best, but we're using their production instance now, so we should just adapt.