fix: parse single-rune Unicode flags

[znnnnnnn-wil]

What type of PR is this?

• bug

What this PR does / why we need it:

• Count Unicode code points, rather than UTF-8 bytes, when selecting the prefix for a flag name.
• Decode the first rune after - before deciding whether an argument is a flag.
• Add a regression test showing that a single-rune Hebrew flag is displayed and parsed like a single-character ASCII flag.

Some code and tests were assisted by Codex. I reviewed, understood, and tested the complete change and take responsibility for it.

Which issue(s) this PR fixes:

Fixes #2433

Special notes for your reviewer:

The parsing change only affects non-ASCII input. ASCII arguments decode to the same rune value as the byte previously inspected.

Testing

• go test . -run '^TestCommand_SingleRuneUnicodeFlag$' -count=1
• go test ./... -count=1
• go run scripts/build.go vet
• go run scripts/build.go check-binary-size
• go run scripts/build.go generate
• go run scripts/build.go gfmrun
• Equivalent public API diff using git diff --no-index --ignore-all-space (no differences)

The repository's go run scripts/build.go test command requires -race; it could not run on this Windows host because CGO is disabled and no C compiler is installed. The same packages passed without -race, and CI will run the race-enabled suite on supported runners.

Release Notes

Fix parsing and help formatting for single-rune Unicode flag names.