chore: promote the 0.6.0 public API surface

[rubenhensen]

Release-time step from CLAUDE.md (release-please does not do this): apply the *REMOVED* entry to PublicAPI.Shipped.txt — NetworkException.Body.get deleted, matching what 0.6.0 actually shipped — and reset PublicAPI.Unshipped.txt to its #nullable enable header.

Verified: dotnet build clean (analyzer accepts the new baseline), tests pass on net10.0.

Part of encryption4all/postguard#247 (workstream D housekeeping).

[dobby-coder]

Thanks, Ruben — I'm on it. I'll review PR #53's PublicAPI baseline changes (the *REMOVED* entry for NetworkException.Body.get in PublicAPI.Shipped.txt and the reset PublicAPI.Unshipped.txt header) against the 0.6.0 release and the workstream D housekeeping context from #247.

[dobby-coder]

Nothing actionable. Verified against the PR head (7d5a491); note the PR was merged as daa98ff while this review was running, so this lands as a post-merge sign-off.

What was checked

• Matches the documented release step. CLAUDE.md ("Tracked public API surface") says: at release time move PublicAPI.Unshipped.txt entries into PublicAPI.Shipped.txt, applying *REMOVED* lines as deletions, and leave the unshipped file with just its #nullable enable header. That is exactly the diff — one deletion in each file, no other edits.
• The removal is real, not just bookkeeping. NetworkException.Body is gone from the source (src/Exceptions/PostGuardException.cs keeps only the private _body field from the #41/#43 info-exposure fix), so deleting the shipped line reflects the surface 0.6.0 actually shipped.
• Timing is right. E4A.PostGuard-v0.6.0 is tagged and the release commit (174dbf8) is on main, so the promotion is post-release, not premature.
• Baseline is enforced, and it passes. RS0016/RS0017 are both in <WarningsAsErrors> in src/E4A.PostGuard.csproj, so a shipped line for a member that no longer exists (or an untracked member) breaks the build — the green "Build and test" check is real evidence the new baseline is accurate, not silently drifting. The remaining NetworkException(int statusCode, string! body, string! url) -> void line still matches the live constructor.
• Nothing else was pending. PublicAPI.Unshipped.txt at the merge base held only the one *REMOVED* entry, so no additions were dropped on the floor. Shipped file stays sorted at 100 lines.

Rule sweep

• Conventional-commit PR title — .github/workflows/pr-title.yml exists on this repo and the check passes.
• Issue linking — Part of encryption4all/postguard#247 is the deliberate non-closing form, fully qualified for cross-repo, and there is no stray closing keyword elsewhere in the body. The tracking issue correctly stayed open on merge.
• Scope and PR body — one commit, two files, no unrelated edits; body is short and states the verification actually run.
• No test change expected here: this is a metadata baseline with no behaviour change, and the analyzer is the test.

One pre-existing thing, out of scope for this PR and not worth its own change: _body is now assigned and never read. That is intentional per the comment above it, and it is left over from #43, not introduced here.

[dobby-coder]

Reviewed and approved — pull request review. Nothing actionable, 0 findings.

Looks like this merged as daa98ff while I was still reviewing, so my approval landed a few minutes post-merge — but it's on the PR either way. I checked it against the CLAUDE.md release step and it's exactly right: one *REMOVED* deletion in PublicAPI.Shipped.txt for NetworkException.Body.get (which is genuinely gone from the source), and PublicAPI.Unshipped.txt reset to just its #nullable enable header. The build enforces RS0016/RS0017 as errors, so the green "Build and test" check is real confirmation the new baseline matches what actually shipped in 0.6.0, not just a tidy-up. Title, cross-repo reference to #247, and diff scope all check out too.