Update how metadata are stored in bonsai input

[mhkc]

This PR makes changes to the output data format.

• RunMetadata has been split into two seperate data classes, one for recording sequencing information and the other for information on the pipeline.
• LIMS id and sample name was moved from RunMetadata into the SampleBase object, which makes more sense.
• Split SV and SNV from tbprofiler output
• Update annotate_delly to extract SVs from vcf
• Fix genome_annotation appending bug
• Fix variant sorting as reference_sequence can be None
• Add indel_variants to json result and ability to filter vcf into various categories