Implement case-folded comparisons as stdlib SPI

[natecook1000]

Following on the behavior improvement in #383, we should provide case-folded comparisons as stdlib SPI. We should be able to skip UTF-8 encoding and do character-by-character canonicalization much more efficiently that way.