Add mavedb logging and fix mem

[ainefairbrother]

This PR addresses ticket ENSVAR_6893.

• The main goal is to add consistent and detailed logging so that it is clear why particular URN IDs may have failed following a run. Previously, IDs were failing without clear logs and it was not easy to understand exactly why without re-running those IDs.
• Secondarily, it makes the metadata extraction (extract_metadata.py) more robust to subtly differing file formats, which can occur as the MaveDB data release has not stabilised yet.
• It also fixes mapping file matching pattern in import_from_files.sh - was too liberal.
• Updatesnextflow.config to get more detailed tracing to understand memory assignation vs usage.
• Updates the memory calculation in variant_recoder.nf to reduce and cap memory requested based on previous run traces.