Consolidate decimal digit -> str functions in tokenizer.c

[WillAyd]

@Alvaro-Kothe implemented a really nice PR in #62658 (comment) with a copy_string_without_char function to remove the separator from a numeric string

We also have a function _str_copy_decimal_str_c at

pandas/pandas/_libs/src/parser/tokenizer.c

Line 1741 in 5eaf80b

static char *_str_copy_decimal_str_c(const char *s, char **endpos, char decimal,

that is similar, but:

• allocates a buffer on the heap
• supports replacement of multiple characters
• reads and writes per-byte

It would be nice to combine both of these functions into one and take advantage of their individual strengths, utlimately getting a function that can replace multiple characters into a caller-provided buffer (ideally on the stack), and which moves blocks of memory instead of individual characters, where applicable.

strtok_s / strtok seem like they could be really useful here, or alternatively strpbrk

[heoh]

I'm interested in this. Would it be okay if I try to contribute?

[heoh]

take