Unicode filenames preserved in wiki slugs

The Memory Wiki plugin was stripping every non-ASCII character from page titles when generating filenames. A title like "大语言模型概述" would silently become "page.md"—meaning all Chinese titles mapped to the same file, with later pages overwriting earlier ones. The regex /[^a-z0-9]+/g accepted only basic Latin letters and digits.

Two functions received Unicode-aware replacements: slugifyWikiSegment() now uses /[^\p{L}\p{N}\p{M}]+/gu, preserving letters from any writing system plus digits and combining marks. The same fix landed in normalizeClaimTextKey() to prevent CJK contradiction notes from collapsing into identical empty-string keys.

To keep filenames within filesystem limits, a new truncation system adds SHA-1 hash suffixes when segments exceed byte thresholds. This handles edge cases where Unicode characters expand byte length beyond platform limits.

The fix lives in the memory-wiki extension, used by bridge and unsafe-local page resolution modules. Changelog entry credits @zhouhe-xydt and @vincentkoc.