Fix recent perf regressions

[bicknellr]

Mitigates perf regressions introduced in some recent PRs by removing the namespaceURI check (externally observable, but never released) and storing unflushed definitions by name only.

[kevinpschaaf]

LGTM, the change in the map code looks equivalent. Somewhat surprising that iterating the array of names shows up as meaningfully faster than a forEach over the map (which iiuc requires a key sort) given the benchmark used for testing wasn't taxing a large number of element tag names; but fine for now and we can do another round of building out benchmarks and micro-opting later.

[sorvell]

LGTM on the code, but let's make sure to test performance on Edge and IE (the main browsers where the polyfill is needed) before landing this.

[bicknellr]

Here's two different runs on Edge that benchmark walking a pseudo-random tree of 8192 nodes with ~50% trivial custom elements. (cf4d955 is before the PRs we were concerned about, f0a44fc is with those PRs, 6f219d4 is with this PR also)

___________________________________________________
_     Version _ <none>                            _
___________________________________________________
_     Browser _ edge                              _
_             _ 18.17763                          _
___________________________________________________
_ Sample size _ 500                               _
___________________________________________________

____________________________________________________________________________________________________________
_ Benchmark      _ Bytes     _          Avg time _ vs #0 (cf4d9550) _ vs #5 (f0a44fc3) _ vs fix (6f219d44) _
____________________________________________________________________________________________________________
_ #0 (cf4d9550)  _ 22.18 KiB _ 34.79ms - 35.02ms _                  _           faster _            faster _
_                _           _                   _         -        _          3% - 4% _           0% - 1% _
_                _           _                   _                  _  1.04ms - 1.37ms _    0.06ms - 0.4ms _
____________________________________________________________________________________________________________
_ #5 (f0a44fc3)  _ 24.81 KiB _ 35.99ms - 36.22ms _           slower _                  _            slower _
_                _           _                   _          3% - 4% _         -        _           2% - 3% _
_                _           _                   _  1.04ms - 1.37ms _                  _   0.80ms - 1.14ms _
____________________________________________________________________________________________________________
_ fix (6f219d44) _ 24.81 KiB _ 35.01ms - 35.26ms _           slower _           faster _                   _
_                _           _                   _          0% - 1% _          2% - 3% _          -        _
_                _           _                   _  0.06ms - 0.40ms _   0.8ms - 1.14ms _                   _
____________________________________________________________________________________________________________

___________________________________________________
_     Version _ <none>                            _
___________________________________________________
_     Browser _ edge                              _
_             _ 18.17763                          _
___________________________________________________
_ Sample size _ 500                               _
___________________________________________________

____________________________________________________________________________________________________________
_ Benchmark      _ Bytes     _          Avg time _ vs #0 (cf4d9550) _ vs #5 (f0a44fc3) _ vs fix (6f219d44) _
____________________________________________________________________________________________________________
_ #0 (cf4d9550)  _ 22.18 KiB _ 34.82ms - 35.01ms _                  _           faster _            faster _
_                _           _                   _         -        _          3% - 4% _           0% - 1% _
_                _           _                   _                  _    1.1ms - 1.4ms _      0ms - 0.29ms _
____________________________________________________________________________________________________________
_ #5 (f0a44fc3)  _ 24.81 KiB _ 36.06ms - 36.28ms _           slower _                  _            slower _
_                _           _                   _          3% - 4% _         -        _           3% - 4% _
_                _           _                   _  1.10ms - 1.40ms _                  _   0.95ms - 1.27ms _
____________________________________________________________________________________________________________
_ fix (6f219d44) _ 24.81 KiB _ 34.95ms - 35.17ms _           slower _           faster _                   _
_                _           _                   _          0% - 1% _          3% - 3% _          -        _
_                _           _                   _  0.00ms - 0.29ms _  0.95ms - 1.27ms _                   _
____________________________________________________________________________________________________________

The relative times between the commits are roughly the same as what I was seeing in Chrome.