test: cover the gaps left by the audit

[MaxMichel2]

Summary

A grab-bag of test coverage gaps identified while auditing the networkmock family, independent of the OpenAPI migration (tracked in the epic, #72). Each bullet below is independently addressable — feel free to split into separate PRs (or separate issues) rather than doing all of them at once; this issue exists to record them so they aren't lost, not to mandate one big PR.

Gaps

1. Delay precedence — Operation.delayMs ?: group.defaultDelayMs resolution (devview-networkmock-core/src/commonMain/kotlin/com/worldline/devview/networkmock/core/repository/MockConfigRepository.kt:366) has no dedicated test asserting the precedence order (operation-level delay wins, falls back to group default, falls back to no delay).
2. Ambiguous-match precedence — if two groups' configured hosts both match an incoming request's hostname, findMatchingMock (repository/MockConfigRepository.kt:304-374) returns whichever is declared first, with no test pinning this down. (If Feature/feature flip #2 has landed by the time this is picked up, this is likely already covered there — check before duplicating.)
3. NetworkMockEndpointScreen has zero tests — no coverage exists for the sticky-header status-family grouping (devview-networkmock/src/commonMain/kotlin/com/worldline/devview/networkmock/NetworkMockEndpointScreen.kt:119-121, :234-278), the "No mock" row (:208-225), or response-row selection (:244-262).
4. Preview/diff bottom sheet has zero tests — NetworkMockEndpointPreviewBottomSheet.kt and the diff rendering pipeline it drives (utils/DiffLineUtils.kt) have no test coverage, despite PreviewSheetState itself (the state machine) being tested in PreviewSheetStateTest.kt.
5. No test loads the real sample mocks.json (or, post-migration, the real sample spec) — every existing config-loading test uses hand-built fixture data (MockTestData.kt), so the shipped sample can silently drift out of sync with what the parser actually accepts. Add a test that loads sample/network/src/commonMain/composeResources/files/networkmocks/mocks.json (or its post-migration equivalent) through the real repository and asserts it parses successfully.
6. queryParams/queryParameters matching through the Ktor plugin — RequestMatcherTest.kt covers matchesQueryParams in isolation thoroughly, but NetworkMockPluginTest.kt (devview-networkmock-ktor/src/androidHostTest/...) has no end-to-end test exercising query-param matching through the actual plugin interception path.
7. No assertion on response headers/content-type in NetworkMockPluginTest.kt — existing tests check status and body but not that Content-Type: application/json is actually present on the synthetic response.

Acceptance criteria

• Each gap above has at least one new test closing it, or is explicitly noted as already covered by a different in-flight PR (to avoid duplicate work — check the epic Epic: OpenAPI migration and environment removal (0.2.0) #72 for what's landed).
• New tests follow this repo's existing conventions: commonTest + JVM for repository/matcher logic (see RequestMatcherTest.kt, MockConfigRepositoryTest.kt for style), androidDeviceTest + fakes for Compose UI (see EndpointCardTest.kt for style), androidHostTest + MockK for the Ktor plugin (see NetworkMockPluginTest.kt for style).

Files likely touched

• devview-networkmock-core/src/commonTest/kotlin/com/worldline/devview/networkmock/core/repository/MockConfigRepositoryTest.kt
• devview-networkmock/src/androidDeviceTest/kotlin/com/worldline/devview/networkmock/... (new test files for the endpoint screen and preview sheet)
• devview-networkmock-ktor/src/androidHostTest/kotlin/com/worldline/devview/networkmock/ktor/plugin/NetworkMockPluginTest.kt

[github-actions]

This issue is stale because it has been open for 21 days with no activity.