fix(zig): count multiline strings and character literals

[LuisFigueroaG]

Two problems in how scc reads Zig strings. Both made comments count as code.

A multiline string line (\\...) ends at the newline, and languages.json says so (end "\n"). But stringState stops at every newline before it looks for a closer, so the closer was never matched and the string ran to the end of the file. The first commit ends a newline-terminated string when its line ends.

Zig uses ' only for character literals, and it wasn't declared as a quote. So in '"' the double quote opened a string that swallowed every line up to the next ". The second commit adds ' as a quote and regenerates processor/constants.go.

On 218 files from Zig's standard library, code goes from 187,635 to 182,618 lines. The result (182,618 code, 24,399 comments, 21,732 blanks) matches a plain line-by-line classification exactly. That comparison is reliable for Zig, which has no block comments and no strings that span lines.

Tests are in processor/workers_zig_test.go: a multiline string followed by comments, a '"' literal followed by comments, and a test that '\'', '\\' and an apostrophe inside a \\ line still count correctly. ./test-all.sh passes.

I explicitly license this contribution under the MIT licence.

[boyter]

TY!