Implement table-driven F.O.S. lesson completion

[ToumanLin]

Problem

F.O.S. lessons and assessments can be entered and played, but valid lesson-result submissions are rejected because the server has no supported lesson chapter-clear rule. This prevents progress from being persisted and can leave later assessments and rewards unavailable.

Summary

• Resolve submitted lessons from the authoritative course-stage table.
• Persist accepted lesson stage results and mark a lesson clear once every configured stage has a saved record.
• Keep lesson completion distinct from maximum points/full-star completion.
• Preserve the configured ClearPoint threshold for exams.
• Preserve existing pending/saved result, downgrade, reward, retry, and BSON persistence behavior.
• Add C1/C2 coverage, including a partial-star final C2 stage.

Evidence and data sources

A controlled check on the official 4.7.0 server confirmed that a lesson chapter becomes clear when its configured stages are complete even when the final Vol is not full-star. This establishes that lesson chapter clear is distinct from full-star/max-point completion.

The authoritative CourseChapter table identifies lessons with StageType=1 and lists their stage IDs; lesson rows have no ClearPoint. Exams use StageType=2 and ClearPoint=9. The inspected course schemas contain no FinishCondition field. The 4.7.0 client consumes the server's chapter IsClear flag while calculating stage-record completion and full-star completion separately.

No captured response, stage-specific value, or player state is used by production code. Captured/observed transitions are test oracles only.

Testing

• dotnet build AscNet/AscNet.csproj
• dotnet run --project AscNet.Test/AscNet.Test.csproj -- --course-only

[reiserFSs]

Reviewed against 67b9b1ec794540da4a805f9b5deb2a0adca54b7f. Hold pending confirmation of the lesson-completion rule.

The project builds, --course-only passes, and GitHub reports a clean merge. No published CI checks were reported.

The actual runtime change in CourseModule.cs:44-53 treats every lesson chapter as having a completion rule and sets it clear once all configured stage IDs have saved records. It does not read the configured lesson FinishCondition/score threshold described in the PR body; those fields are not present in the inspected CourseStage schema.

The client consumes the server's chapter IsClear flag. Its stage-record presence checks and distinction between completion and maximum points do not independently establish the proposed chapter-completion formula. The added C1/C2 tests exercise the chosen formula using full-star results, rather than providing an independent oracle for it.

Please provide authoritative source or a controlled retail transition establishing the rule, and align the description/tests with that evidence. Alternatively, this needs explicit maintainer approval as a documented local policy rather than a claim of established parity.

To be clear: I have not established that a zero-star result must be rejected or that lessons require full stars. The blocker is the missing authority for the formerly unsupported chapter-clear rule, not a proven score-threshold defect. No changes were pushed during review.

[ToumanLin]

Thanks — your review is correct. The original PR description overstated the available table data: the inspected course schemas do not contain a FinishCondition, and lesson rows do not define a score threshold. I have removed that claim from the intended description.

I also performed a controlled check against the official 4.7.0 server:

• All configured Vols in a lesson were completed.
• At least one optional star was deliberately left unearned.
• The client still showed the lesson chapter as clear.

This confirms that lesson chapter completion does not require full stars or the maximum lesson-point total.

That retail observation is consistent with the authoritative tables and client behavior:

• Lesson chapters are identified by StageType=1 and their configured StageIds; they have no ClearPoint.
• Exams use StageType=2 and the configured ClearPoint=9.
• The client considers a stage complete when its saved stage record exists.
• Chapter IsClear is consumed from the server, while full-star completion is calculated separately.

I pushed follow-up commit 5c811a780786b438e362793fe984c6f12b35c407. The C2 test now saves the final lesson stage with only one star, verifies that its total remains below the full-star maximum, and verifies that IsClear becomes true only after all configured stage records exist.

Validation after the update:

• dotnet build AscNet/AscNet.csproj
• dotnet run --project AscNet.Test/AscNet.Test.csproj -- --course-only

Both pass. Production code remains table-driven and does not use captured response values or player-state fixtures.