Bugfixes, and one fast report

[MattWellie]

Fixes

Miah has spotted a couple of bugs

• the arguments to the PED -> PhenoPacket converter need correction
• the HTML generator script doesn't tolerate an output path outside of the current directory

Proposed Changes

• correction to the ped conversion arguments
• use os.makedirs based on the target location, not the current location

Report Fixes

• for the CPG pipeline, we need to create the main document: The CPG pipeline will only register outputs in metamist if the exact file has been created, so we always make that, then optionally do splitting.
• the jinja template engine is now started up with trim_blocks and lstrip_blocks - these arguments stop jinja from inserting empty lines in the HTMLs we produce. These account for over half the file length in lines, but their removal didn't speed anything up
• I've also fixed the clinvar hyperlink - I was creating a hyperlink with no content if the allele doesn't have a clinvar entry. This removal also didn't meaningfully speed things up, but it's silly.
• This includes a complete GPT-4 inspired revolution on the report - now it's paginated, with a spectacular performance upgrade.
• Sample/Family rows are now collapsible using this functionality

still required

• now that we don't need to split report, we can try to reconcile this new jquery with the original child table functionality

Checklist

• [ ] Related Issue created
• [ ] Tests covering new change
• [x] Linting checks pass

[MattWellie]

something like this would be useful to combine the child-row details with the main view, but DataTables seems to be less straightforward than it was using TableSorter e.g. https://datatables.net/dev/accessibility/DataTables_1_10/examples/api/row_details.html ?