If no range is specified, they will return the same number.
But if different ranges (optionally with different offsets) are specified, different numbers should be returned due to the different modulo and/or offsets of the ranges.
Thus, multiple 'H' with ranges in one field should be allowed for convenience.

Easier to see which test failed

Useful for putting the parsed H fields back together

… integers

The issue was the missed `free()` of the result fields being merged into the final field.
As the original subfields are already freed in `replace_hashed()`, this should now happen in the final free for the list subfields.
Also removed the `str_append()` utility, as allocating one big string and filling it with the single results (via sprintf) may take more runtime (for looping twice), but simplifies memory management and improves readability.

Contents are identical to subfields. Can simply use it instead

It requires tracking of the substring lens, but it should be more efficient with shorter writes.

Regarding parsability, combinability with other options

Add commas when needed. Also use strcpy and a tracking pointer to make copying more efficient

Free the accum_field and subfields always if an error occurs during field replacement (in a list) and they were malloced