Benchmarking via AirspeedVelocity

[henry2004y]

Test if benchmarking with AirspeedVelocity.jl gives better workflow compared with https://github.com/tkf/BenchmarkCI.jl.

Features that I enjoy:

1. Benchmark results are updated instead of posting a new one after each new push.
2. Result table is concise

Feature I wish to have:

1. Memory allocation check (not the default, but can be turned on with --mode=time,memory).
2. Summary of regression tests (used to be there in PkgBenchmark.jl).
3. Better looking output table.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 84.39%. Comparing base (6cd409b) to head (b403a17).
Report is 1 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master     #206   +/-   ##
=======================================
  Coverage   84.39%   84.39%           
=======================================
  Files           9        9           
  Lines         692      692           
=======================================
  Hits          584      584           
  Misses        108      108           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[github-actions]

Benchmark Results

	master	b403a17...	master/b403a17340af4f...
trace/GC/in place	0.145 ± 0.044 ms	0.149 ± 0.045 ms	0.97
trace/analytic field/in place	6.5 ± 3.2 μs	7.86 ± 3.3 μs	0.827
trace/analytic field/in place relativistic	9.58 ± 3.3 μs	9.36 ± 3.3 μs	1.02
trace/analytic field/out of place	6.06 ± 2.6 μs	6.03 ± 2.6 μs	1.01
trace/numerical field/Boris	2.23 ± 0.16 μs	2.23 ± 0.16 μs	1
trace/numerical field/Boris ensemble	4.3 ± 0.96 μs	4.31 ± 0.97 μs	0.997
trace/numerical field/in place	20.4 ± 3.2 μs	20.4 ± 3.3 μs	0.997
trace/numerical field/out of place	15.1 ± 2.2 μs	15.3 ± 2.3 μs	0.989
trace/time-dependent field/in place	10 ± 3.6 μs	8.37 ± 3.5 μs	1.19
trace/time-dependent field/out of place	7.2 ± 2.9 μs	6.85 ± 2.9 μs	1.05
time_to_load	2 ± 0.0035 s	2 ± 0.007 s	0.999