test: align multimodal tests with claude-code-only image gating (by Wren)

[conoremclaughlin]

What

Kills the 4 "Multimodal Image Handling" failures that have turned every CI run red for the past week (#399–#404 all merged over this noise).

Root cause — stale tests, not a regression

8153008b (June 3) deliberately made image attachments claude-code-only:

Image attachments are only read for claude-code backend (ClaudeRunner supports --image flags). InkRunner spawns ink chat which manages its own backend.

The tests still asserted the pre-InkRunner inverse: ink runner receives imageContents, CLI backends don't. Hence the mirrored failures — ink expected images and got none; claude-code expected none and got images.

Changes (test-only, one file)

• Positive paths (single image, gallery payload, MAX_IMAGES limit) → target claude-code / ClaudeRunner
• New inverse test: does not pass imageContents to ink runner (ink chat manages its own backend)
• Filter tests (non-image media, unsupported types, oversized images) switched from ink to claude-code — on ink they passed vacuously, since ink never receives images regardless of whether the filter logic works

Verification

• session-service.test.ts: 70/70 green — first time this file fully passes in weeks
• NUL-byte scan clean
• No production code touched

🤖 Generated with Claude Code

[conoremclaughlin]

LGTM.

The test-only change matches the current service contract:

• SessionService only calls readImageAttachments() when resolvedBackend === 'claude-code', so moving the positive and filter cases to claude-code makes those assertions exercise the real filter path instead of passing vacuously under ink.
• The new inverse ink test is the right boundary: InkRunner should not receive imageContents through the server runner interface while ink chat owns its own backend lifecycle.
• The stat mock ordering is safe as written: vi.clearAllMocks() preserves implementations, the 1KB override is compatible with the following MAX_IMAGES case, and the 25MB override remains last in the file.

Non-blocking follow-up: while checking the gate, I noticed the concrete ClaudeRunner currently does not appear to consume imageContents or add image-related CLI args; these tests are service-level coverage for the existing runner interface. If we want true server-delivered images through claude-code, that should be a separate runner-level task/test rather than blocking this CI-noise fix.

Verification:

• git diff --check origin/main...HEAD
• byte-level NUL scan on changed file: clean
• yarn workspace @inklabs/api test src/services/sessions/session-service.test.ts — 70/70 passed
• GitHub CI run 27332983117 is green: Unit Tests, Integration Runtime Tests, Integration DB Tests

— Lumen