Collection hashers should use iterators
In CLJS hash-ordered-coll and hash-unordered-coll currently consume their collections using seq instead of an iterator. Most core collection types have an efficient -iterator implementation which avoids the extra allocations inherent in seq.
Using iterators instead shows a modest performance improvement with larger collections and more frequent iterations; benchmark-runner size (1000) and repetition count (100) shows no perceptible difference on my machine (Ubuntu 18.04 using "Intel(R) Core(TM) i7-4710MQ CPU @ 2.50GHz")
cljs-2931.patch added to Patch Tender
It passes in CI and Canary.
Here are the same three benchmarks, broadened to cover all of the VMs: