Add segment-to-JSONL export tool for research access

[erskingardner]

Problem

Pensieve segment files are the canonical archive, but they are stored as length-prefixed notepack payloads, optionally wrapped in gzip. That is efficient for storage and ingestion, but not convenient for academics or other researchers who expect line-delimited Nostr JSON events.

Why this matters

Researchers should not need to understand Pensieve internals or notepack framing just to use an archive snapshot. A JSONL export tool gives them a universal, inspectable format while preserving notepack as the canonical storage format.

Suggested implementation

Add a CLI tool, likely under crates/pensieve-ingest/src/bin/, such as segment-to-jsonl.

The tool should:

• Accept one segment file or a directory of segments.
• Read both .notepack and .notepack.gz files.
• Decode each [u32 little-endian length][notepack bytes] record.
• Emit one canonical Nostr JSON event per output line.
• Preserve deterministic ordering by segment filename and record order.
• Surface parse/truncation errors with file and byte offset context.

Acceptance criteria

• A .notepack segment exports to valid JSONL.
• A .notepack.gz segment exports to valid JSONL.
• Output lines contain valid Nostr event fields: id, pubkey, created_at, kind, tags, content, sig.
• Directory input processes segments in lexical filename order.
• Tests cover uncompressed, gzip-compressed, and truncated segment cases.
• just precommit passes before merging.