perf: optimize body parsing and buffering in request handling with onDataV2

[nigrosimone]

Switches from onData to onDataV2 which exposes maxRemainingBodyLength the exact number of bytes remaining in the request body. This enables several performance optimizations like pre-allocated body buffer.

PR BENCHMARK

Test	Express req/sec	uExpress req/sec	Express throughput	uExpress throughput	uExpress speedup
engines/art	4.32k	10.19k	1.89 MB/sec	4.48 MB/sec	2.37x
middlewares/body-urlencoded	7.15k	33.86k	1.11 MB/sec	5.30 MB/sec	4.77x
middlewares/compression-file	4.76k	6.97k	2.17 MB/sec	3.18 MB/sec	1.47x
middlewares/express-static	2.42k	6.89k	592.86 MB/sec	1.64 GB/sec	2.83x
routing/hello-world	11.07k	70.89k	1.84 MB/sec	11.83 MB/sec	6.43x
routing/middlewares-100	10.14k	14.65k	1.59 MB/sec	2.31 MB/sec	1.45x
routing/nested-routers	10.18k	59.03k	1.63 MB/sec	9.51 MB/sec	5.83x
routing/routes-1000	3.27k	54.06k	523.02 KB/sec	8.51 MB/sec	16.66x
streaming/readable-hash-4mb	10.82k	69.21k	4.00 MB/sec	25.67 MB/sec	6.42x
streaming/writable-no-content-length	500.09	533.87	2.44 GB/sec	2.62 GB/sec	1.07x
streaming/writable-with-content-length	513.71	417.06	2.51 GB/sec	2.05 GB/sec	0.82x

MASTER BENCHMARK

Test	Express req/sec	uExpress req/sec	Express throughput	uExpress throughput	uExpress speedup
engines/art	4.06k	7.86k	1.78 MB/sec	3.46 MB/sec	1.94x
middlewares/body-urlencoded	5.84k	18.26k	931.84 KB/sec	2.86 MB/sec	3.14x
middlewares/compression-file	3.26k	6.72k	1.49 MB/sec	3.07 MB/sec	2.06x
middlewares/express-static	2.01k	7.19k	492.15 MB/sec	1.72 GB/sec	3.58x
routing/hello-world	7.74k	52.99k	1.28 MB/sec	8.84 MB/sec	6.91x
routing/middlewares-100	7.54k	12.55k	1.18 MB/sec	1.97 MB/sec	1.67x
routing/nested-routers	7.29k	42.18k	1.17 MB/sec	6.80 MB/sec	5.81x
routing/routes-1000	3.41k	38.51k	545.72 KB/sec	6.06 MB/sec	11.37x
streaming/readable-hash-4mb	7.87k	53.41k	2.91 MB/sec	19.82 MB/sec	6.81x
streaming/writable-no-content-length	542.93	581.82	2.66 GB/sec	2.85 GB/sec	1.07x
streaming/writable-with-content-length	560.60	627.22	2.74 GB/sec	3.07 GB/sec	1.12x

[uNetworkingAB]

You can only use maxRemainingBodyLength if it is less than your upper buffer size. maxRemainingBodyLength will be essentially infinity for chunked requests.

[nigrosimone]

Good catch @uNetworkingAB . I was using maxRemainingBodyLength for preallocation too broadly, but that is not valid for chunked requests. I changed it so it is only used when it is safely bounded by the parser limit; otherwise the code falls back to dynamic buffering. You can make a new code review?