feat(Crypto): add ULID support

[Cyberistic]

What is the problem this feature would solve?

Add readonly randomULID to interface Crypto

ULIDs are better than UUIDs in some cases. (shorter, lex-sortable, no hyphens, etc)
ULIDs are also case-insensitive, contains no confusing characters (like O/0 or I/1), and is easily double-click selectable in terminal windows or URLs (again, no hyphens!).

https://github.com/ulid/javascript

Thank you :)

[Cyberistic]

duckdb supports ULID btw: https://duckdb.org/community_extensions/extensions/ulid

[gold-beyond]

Thanks for the write-up. This one seemed worth doing, and most of the machinery was already in place: Crypto has randomUUIDv4 / randomUUIDv7, and Schema.isULID already exists for validation, so what was missing was the encoder.

Opened #8233.

A few notes from writing it:

• The 128 bits are left-padded with two zero bits before being split into 26 five-bit groups, so the first character carries only three bits (7 is the largest leading character). Doing that padding at the end still produces valid-looking ULIDs but breaks their sort order, which is why the tests pin the exact encoding instead of just the length.
• I cross-checked against ulid@3.0.2 (the ulid/javascript package) over 500 random timestamp/randomness pairs: identical bytes in, identical 26 characters out, and the time prefix matches the README example 01ARYZ6S41 for 1469918176385.
• Two ULIDs generated in the same millisecond are ordered by their random characters only. Strictly increasing IDs need the monotonic factory from the spec; I left that out so this stays a drop-in addition, but it's easy to add if you'd rather have it.

Whatever randomULID returns validates with Schema.isULID.