Provide special parser for original `Crockford's base32` possibility

[kachick]

ref: #57 and ulid/spec#57

My suggestion around the issue is written in oklog/ulid#69 (comment) and https://github.com/ulid/spec/pull/57/files#diff-b335630551682c19a781afebcf4d07bf978fb1f8ac04c6bf87428ed5106870f5R31

All characters except 0123456789ABCDEFGHJKMNPQRSTVWXYZabcdefghjkmnpqrstvwxyz are invalid in ULID

I think ignoring them is the desirable spec for actual use-case, rather than strict following original Crockford's base32.

But providing the parser for covering the original Crockford's base32 might be useful for exists difficult ULIDs.

How

• When provide to it, the parser should have another name with current one, I feel 🤔

[kachick]

When provide to it, the parser should have another name with current one, I feel 🤔

Or adding keyword arguments as an optional parameter? 🤔

[kachick]

Just providing new parser is not a hard task. Considering to update or keep exists methods is annoy 🤔

• ULID.valid?
• ULID.scan
• ULID#===
• ULID#pattern
• ULID#strict_pattern