[P1] sync.py downloads data and then drops it (updated_at, sources, tags, collection labels, hub metrics)

[michellzappa]

Problem

scripts/sync.py downloads more than it writes out:

• updated_at is selected from Supabase but never copied into the record built by build_enriched(), so it is null in all rows of indexes/technologies.json.
• Sources: technologies_sources is downloaded and never written. The CMS has also moved sources into technology_evidence, so the legacy table may be out of date.
• Tags: per-technology tag1/2/3 are only exported as per-hub counts in indexes/tags.json, not on each entry.
• collection_label isn't exported, so cities collections show up as IDs like M7CFmLD9Qx2KxloytEYe6w.
• Hub fields about, indicator_explainers and metrics_config are downloaded but not written to indexes/hubs.json, which causes [P0] trl / impact / investment are mislabelled for 9 hubs (986 entries) #5.
• Hubs with no published technologies (e.g. lexicon) are listed with a count of 0.

Fix

• Pass updated_at through in build_enriched(). While there, fix the misindented "sources" line.
• Read sources from technology_evidence, falling back to technologies_sources.
• Add collection_label and tags to the frontmatter, and sources to the body as a ## Sources list.
• Add about and metrics_config to hubs.json.
• Leave out hubs whose research row is unpublished or that have 0 published technologies.
• Add a unit test in scripts/test_sync.py for each of these.

Runbook: docs/improvements/03-schema-v2.md