test: e2e coverage for PR #35 (default search mode) and PR #36 (summary removed)

[davidgut1982]

Summary

Adds tests/e2e/test_pr35_pr36.py with 4 tests that lock in the behaviour introduced by PRs #35 and #36.

PR #35 — default search mode is now hybrid (not hardcoded fts)

handle_kb_search in server.py now reads LORE_SEARCH_MODE_DEFAULT (default "hybrid") instead of hardcoding "fts". The response includes requested_mode (intent, pre-degradation) and search_mode (executed mode).

Tests added (TestDefaultSearchMode):

• test_no_mode_requests_hybrid — no mode arg → requested_mode == "hybrid"
• test_explicit_fts_requests_fts — search_mode="fts" → requested_mode == "fts"
• test_explicit_hybrid_requests_hybrid — search_mode="hybrid" → requested_mode == "hybrid"

Note: assertions key off requested_mode not search_mode — staging has semantic search disabled so search_mode legitimately degrades to fts; requested_mode captures the caller's intent.

PR #36 — "summary" removed from search mode enum

Literal["summary"] was dropped from server_fastmcp.py's kb_search signature. Passing it must now surface as an MCP validation error.

Tests added (TestSummaryModeRemoved):

• test_summary_mode_raises — search_mode="summary" raises LoreClientError

Test results

Verified against lore-staging (CT 200, http://localhost:5556):

tests/e2e/test_pr35_pr36.py   4 passed
tests/e2e/test_kb_lifecycle.py  13 passed
======================== 17 passed in 13.85s ========================

🤖 Generated with Claude Code