Test UTF-8 identifier validation.

[sunfishcode]

This adds tests to the utf8 branch for validating import module and field names, and custom section names.

This supersedes #450

• add more tests for overlong encodings
• add more tests for out-of-range code points
• add more tests for UTF16 surrogate halves
• add tests for obsolete 5-byte and 6-byte encodings
• add tests for UTF16 and UTF32 BOMs
• convert tests for invalid strings to use the binary format syntax so that decoding happens separately from parsing
• add comments
• test that a UTF-8 BOM is not rejected

[rossberg]

Thanks! LGTM (reviewed offline, seems like GitHub can't handle UTF-8).

[rossberg]

Argh, sorry I noticed too late that this is a PR against the utf8 branch, which I just landed. So you probably have to merge to master manually.

[sunfishcode]

Github can handle UTF-8, but whoever created those files somehow left invalid byte sequences all over the place ;-). I've now changed the comments to escape those byte sequences, so that the files can be displayed without issue.

It looks like I can't change the target branch for this PR, so I'll create a new PR and merge that, with your LGTM here.