feat: add Warp agent support

[mariow]

Add support for parsing Warp (Oz) agent sessions from its local SQLite database. Warp stores conversation data in warp.sqlite at platform- specific paths, similar to how OpenCode uses a DB-backed approach.

New files:

• internal/parser/warp.go: Parser for Warp's agent_conversations and ai_queries tables. Extracts user messages, project info from working directories, token usage, and synthesizes tool call messages from aggregate tool usage stats in conversation metadata.
• internal/parser/warp_paths.go: Platform-specific default directories (macOS, Linux, Windows).
• internal/parser/warp_test.go: Tests with SQLite fixture covering session discovery, parsing, empty/missing DB handling, timestamp parsing, query text extraction, and tool message synthesis.

Modified files:

• internal/parser/types.go: Add AgentWarp type and registry entry (FileBased: false, env var WARP_DIR, ID prefix "warp:").
• internal/parser/taxonomy.go: Add Warp tool name -> category mappings (run_command->Bash, read_files->Read, apply_file_diff->Edit, etc.).
• internal/sync/engine.go: Add syncWarp/syncOneWarp/syncSingleWarp methods mirroring the OpenCode sync pattern, called after OpenCode in the main sync loop.

[roborev-ci]

roborev: Combined Review (cc5d0fd)

Verdict: No medium-or-higher findings; the patch looks clean.

No Medium, High, or Critical issues were identified across the review outputs.

Security review specifically found no security-impacting issues in the patch.

---

Synthesized from 3 reviews (agents: codex, gemini | types: default, security)

[roborev-ci]

roborev: Combined Review (cd4bc80)

Verdict: No Medium, High, or Critical findings identified.

All available review output for cd4bc80 came back clean at the requested severity threshold. The only substantive review found no security-impacting issues, and there were no reported Medium, High, or Critical findings to include.

---

Synthesized from 3 reviews (agents: codex, gemini | types: default, security)

[wesm]

LGTM, thanks!