[GnoVM] Inconsistent testing times on the same code

[leohhhn]

Description

When running gno test . -v on code, for example in the r/demo/counter directory, test times differ between executions:

r/demo/counter on  master [$] 
❯ gno test . -v      
=== RUN   TestIncrement
--- PASS: TestIncrement (0.00s)
=== RUN   TestRender
--- PASS: TestRender (0.00s)
ok      .       0.84s

r/demo/counter on  master [$] 
❯ gno test . -v
=== RUN   TestIncrement
--- PASS: TestIncrement (0.00s)
=== RUN   TestRender
--- PASS: TestRender (0.00s)
ok      .       0.81s

r/demo/counter on  master [$] 
❯ gno test . -v
=== RUN   TestIncrement
--- PASS: TestIncrement (0.00s)
=== RUN   TestRender
--- PASS: TestRender (0.00s)
ok      .       0.82s

Not sure why this is happening, or if it should happen, especially since there is not parallelism or concurrency involved.

Anyways, this might be replaced by the amount of gas consumed instead, which is a more useful metric: #2467

Any ideas @mvertes @thehowl?

[thehowl]

Most of the computation time goes into loading standard libraries and other packages required by the package being tested. The time will be inconsistent due to varying factors on the machine, like disk access times, so there will be some small differences; but as exist in go and as in most computation.

Ideally, we'd separate the test execution time from the time to load the imports; this could be done, and could make the numbers add up correctly.