Prepare 4.12.1 release

[ewrogers]

What changed

• Ignore non-ASCII bytes while decoding spell prompt buffers from compact and pane spellbook snapshots.
• Keep spell name decoding strict so invalid names still reject the snapshot.
• Add deterministic coverage for prompt filtering and strict spell name validation.
• Prepare the changelog, application metadata, and release instructions for SleepHunter 4.12.1.

Why

The client spell prompt buffer can contain high-byte values for a specific character. SleepHunter 4.12.0 decoded this unused prompt field with strict ASCII validation, so one high byte invalidated the entire Spellbook snapshot and left the character unknown. Reinstalling could not help because the data came from live client memory, not a SleepHunter disk cache.

User impact

Affected characters can attach and capture normally. Non-ASCII prompt bytes are discarded, while valid ASCII prompt text and spell names are preserved.

Validation

• dotnet restore SleepHunter.sln
• dotnet build SleepHunter.sln --configuration Release --no-restore
• dotnet test SleepHunter.sln --configuration Release --no-build
• 816 tests passed
• 0 build warnings and 0 build errors
• Release metadata and changelog version validation passed

A manual check against the affected character on the live 7.41 client is still recommended.