remove unneed hashcount, add xxh3 8.33% faster

[delaneyj]

Per the name of the branch. Initially tried to incorporate roaring bitmaps. Was about 30% slower. This is down to matching the current code. If I did batch compares it might be faster but would look much different.

However, removed HashCount. In general use range as it allows the compiler to make optimizations.
Switch from FNV much higher quality (and faster xxh3).

goos: linux
goarch: amd64
pkg: github.com/starskey-io/starskey/bloomfilter
cpu: AMD Ryzen 9 6900HX with Radeon Graphics        
            │   old.txt   │               new.txt               │
            │   sec/op    │   sec/op     vs base                │
Add-16        69.74n ± 0%   66.99n ± 1%   -3.94% (p=0.000 n=25)
Contains-16   46.60n ± 1%   40.77n ± 1%  -12.51% (p=0.000 n=25)
geomean       57.01n        52.26n        -8.33%

            │   old.txt    │               new.txt               │
            │     B/op     │    B/op     vs base                 │
Add-16        0.000 ± 0%     0.000 ± 0%       ~ (p=1.000 n=25) ¹
Contains-16   0.000 ± 0%     0.000 ± 0%       ~ (p=1.000 n=25) ¹
geomean                  ²               +0.00%                ²
¹ all samples are equal
² summaries must be >0 to compute geomean

            │   old.txt    │               new.txt               │
            │  allocs/op   │ allocs/op   vs base                 │
Add-16        0.000 ± 0%     0.000 ± 0%       ~ (p=1.000 n=25) ¹
Contains-16   0.000 ± 0%     0.000 ± 0%       ~ (p=1.000 n=25) ¹
geomean                  ²               +0.00%                ²
¹ all samples are equal
² summaries must be >0 to compute geomean

[guycipher]

Very good work @delaneyj .