Handle all 1-byte codepoints in `tb_iswprint_ex` fast path

[adsr]

Perf boost to avoid binary search on all 1-byte codepoints. Previously it was just for printable ASCII.

I have a implementation that uses a tiny lookup table and bit math and less conditionals. I may update the PR to include that, but I don't expect it to be worth the code complexity.

$ cat test.c

int main(int argc, char **argv) {
    int i, n;
    uint32_t ch;
    uint64_t count;

    n = argc >= 2 ? atoi(argv[1]) : 1;

    for (i = 0; i < n; i++) {
        for (ch = 0; ch <= 0xff; ch++) {
           count += tb_wcwidth(ch);
           count += tb_iswprint(ch);
        }
    }

    printf("count=%zu\n", count);

    return 0;
}
$
$ # compile test-head and test-patch with -O2
$
$ time for i in $(seq 1 10); do ./test-head 100000; done
count=31900000
count=31900000
count=31900000
count=31900000
count=31900000
count=31900000
count=31900000
count=31900000
count=31900000
count=31900000

real    0m3.793s
user    0m3.773s
sys    0m0.020s
$ time for i in $(seq 1 10); do ./test-patch 100000; done
count=31900000
count=31900000
count=31900000
count=31900000
count=31900000
count=31900000
count=31900000
count=31900000
count=31900000
count=31900000

real    0m0.287s
user    0m0.269s
sys    0m0.019s