Using exclusive mode prefetch

[mjp41]

The prefetching is always used to move the cache line to the current core for writing. This change makes it use exclusive mode prefetch and enables it as a feature flag for x64.

[mjp41]

Most benchmarks don't show a difference in mimalloc-bench, but which is heavily producer consumer is around 5% faster.

xmalloc-testN sn-prefetchr 0.202 266232 190.71 38.13 4 36155
xmalloc-testN sn-prefetchw 0.188 278588 192.37 37.22 6 32462
xmalloc-testN sn-prefetchr 0.202 283372 191.57 42.99 21 37615
xmalloc-testN sn-prefetchw 0.189 258012 186.98 42.39 41 39832
xmalloc-testN sn-prefetchr 0.199 276532 194.81 39.12 54 42775
xmalloc-testN sn-prefetchw 0.186 275996 192.95 38.01 0 38266
xmalloc-testN sn-prefetchr 0.201 282280 190.72 38.75 7 42319
xmalloc-testN sn-prefetchw 0.188 276616 189.45 40.59 55 38771
xmalloc-testN sn-prefetchr 0.201 249948 189.32 40.96 0 39287
xmalloc-testN sn-prefetchw 0.186 274604 195.47 34.73 8 36765
xmalloc-testN sn-prefetchr 0.202 272520 190.24 41.92 34 38657
xmalloc-testN sn-prefetchw 0.186 283648 194.17 35.73 46 42787
xmalloc-testN sn-prefetchr 0.202 221424 184.45 49.19 0 43727
xmalloc-testN sn-prefetchw 0.189 274420 190.00 40.81 1 34877
xmalloc-testN sn-prefetchr 0.202 268268 191.08 39.68 42 39695
xmalloc-testN sn-prefetchw 0.188 274436 192.79 35.84 0 35973
xmalloc-testN sn-prefetchr 0.202 254872 187.74 42.88 28 39420
xmalloc-testN sn-prefetchw 0.187 276372 191.79 38.79 18 38002
xmalloc-testN sn-prefetchr 0.201 274560 191.95 41.81 23 34473
xmalloc-testN sn-prefetchw 0.188 285448 192.93 38.21 0 38145

[nwf-msr]

LGTM; thanks for taking the time to chat about this a few days ago