Implement pipelining

[bgrainger]

Fixes #1088

Local (Windows / Docker)

Before

Method	Mean	Error	StdDev	StdErr	Median	Min	Q1	Q3	Max	Op/s	Allocated
OpenFromPoolAsync	1.376 ms	0.0229 ms	0.0214 ms	0.0055 ms	1.372 ms	1.349 ms	1.362 ms	1.385 ms	1.420 ms	726.5	3,411 B
OpenFromPoolSync	1.399 ms	0.0337 ms	0.0966 ms	0.0099 ms	1.354 ms	1.296 ms	1.332 ms	1.453 ms	1.666 ms	714.9	363 B

After

Method	Mean	Error	StdDev	StdErr	Min	Q1	Median	Q3	Max	Op/s	Allocated
OpenFromPoolAsync	746.2 μs	14.79 μs	16.44 μs	3.77 μs	713.0 μs	735.4 μs	748.8 μs	759.5 μs	775.9 μs	1,340.0	3,376 B
OpenFromPoolSync	734.0 μs	13.13 μs	24.98 μs	3.72 μs	695.1 μs	713.3 μs	728.3 μs	747.4 μs	783.1 μs	1,362.4	361 B

Azure (Linux)

Before

Method	Mean	Error	StdDev	StdErr	Min	Q1	Median	Q3	Max	Op/s	Allocated
OpenFromPoolAsync	1.051 ms	0.0161 ms	0.0143 ms	0.0038 ms	1.036 ms	1.039 ms	1.049 ms	1.057 ms	1.083 ms	951.0	4,067 B
OpenFromPoolSync	1.042 ms	0.0206 ms	0.0193 ms	0.0050 ms	1.014 ms	1.028 ms	1.040 ms	1.054 ms	1.079 ms	959.6	363 B

After

Method	Mean	Error	StdDev	StdErr	Min	Q1	Median	Q3	Max	Op/s	Allocated
OpenFromPoolAsync	534.3 us	9.24 us	8.19 us	2.19 us	522.0 us	528.7 us	532.4 us	538.5 us	549.6 us	1,871.7	2,767 B
OpenFromPoolSync	504.7 us	9.44 us	8.83 us	2.28 us	491.8 us	499.0 us	503.0 us	509.4 us	519.2 us	1,981.2	361 B

[bgrainger]

Added in 2.1.0.

[chillitom]

is there somewhere we can read more about the AWS Aurora pipelining issue?

[bgrainger]

Yes: #486 (comment)