[JSC] Add B3 Mul strength-reduction for Float
Created attachment 376629 [details] Patch
This does not preserve signaling NaN. So this optimization is wrong. And we have this optimization for double, and it is also wrong. But this optimization is OK if the B3 comes from JS, since we ensured that JS does not produce signaling NaN.
Comment on attachment 376629 [details] Patch Attachment 376629 [details] did not pass jsc-ews (mac): Output: https://webkit-queues.webkit.org/results/12934628 New failing tests: wasm.yaml/wasm/spec-tests/float_exprs.wast.js.wasm-no-air