[P2] Dataset schema v2: hub metrics, provenance, evidence level, shared IDs across hubs

[michellzappa]

Goal

Make the dataset self-describing and citable. This builds on #5 and #7.

Proposed additions

field	where	purpose
schema_version: 2	run-manifest and every entry	lets downstream users detect the change
metric1/2/3 + metrics_config	entry + hubs.json	replaces mislabelled trl/impact/investment (#5)
created_at, updated_at, last_reviewed	entry	shows how current an entry is
language	hub + entry	moradia is pt-BR, the rest English
collection_label, tags	entry	readable categories
sources[]	entry	evidence (currently absent everywhere)
evidence_level	entry	established / emerging / speculative (#11)
entity_id	entry	links the same technology across hubs (#10)

Also:

• Publish indexes/technologies.jsonl (one full record per line) for search and AI use.
• Publish a CHANGELOG.md per release ([P2] Update cadence, review dates and versioned releases #13).

Migration

Keep v1 field names for one release with a deprecation note in the README, then remove them.

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