feat!(ingest, prepro): replace fasta filter by a specific metadata filter and add ANY aligned option for segmented organisms

[anna-parker]

resolves #

preview URL: https://ingest-nextclade-sort.loculus.org/

BREAKING CHANGES

This only affects segmented organisms, segmented organisms typically use nextclade_sort or nextclade_align for segment assignment. The arguments relating to these two options should now be placed in a dictionary with that heading e.g.

nextclade_dataset_name: nextstrain/cchfv/linked
nextclade_dataset_server: https://raw.githubusercontent.com/nextstrain/nextclade_data/cornelius-cchfv/data_output

becomes

nextclade_align:
  nextclade_dataset_name: nextstrain/cchfv/linked
  nextclade_dataset_server: https://raw.githubusercontent.com/nextstrain/nextclade_data/cornelius-cchfv/data_output

Summary

1. This PR replaces the previous filter rule that filtered sequences by fasta header with a more exact filter that is applied to the metadata fields post grouping.
2. This PR also updates preprocessing to accept multi-segmented sequences where any of the segments have aligned. Currently we only accept sequences where all segments have aligned. For example in a L, M, S segmented organism all of the segments (that were uploaded) must align for the submission to be successful, e.g. if I upload L and M both must align. Now I require only 1 segment to align, the other will be shown on the webpage in this case with no alignment. For 8-segmented organisms such as influenza this is very important as a 8-segmented sequence could error even if only 1 segment did not align, meaning that we previously bad to perform alignment twice - once during ingest and once during prepro.
3. This makes the ingest config clearer by grouping the nextclade_align and nextclade_sort parameters together, each in one dictionary.