Support reading fields with non-ASCII names

[rui-mo]

Description

Currently the tokenizer only considers ASCII characters, and to read fields with non-ASCII names results in an exception with the message 'Invalid subfield path'.

A reproducible unit test is available at rui-mo@8dc1ea4. To make this test work, a temporary change was made to the 'isUnquotedSubscriptCharacter' function by removing 'isalnum'.

To fix this issue, shall we support UTF-8 characters in the Tokenizer, for example, by replacing 'isalnum' with 'u_isalnum' in the ICU library?

[Yuhta]

Gluten should parse the field names into Subfield directly without using the Tokenizer (which we will rename into PrestoTokenizer).

[rui-mo]

@Yuhta Thanks for your feedback. I will take a look.

[rui-mo]

This issue can be solved with the solution recommended above. Thanks.