Improve executor/scenario configs

[na--]

As seen in #1419, sometimes it would be ideal to pre-calculate some execution segment sequences and reuse them, both in the executor/scenario description, as well as in the actual test execution. Now we don't have any mechanism to do that, so we probably should add some hook that gets called before (or maybe after) the validation.

Another issue that would probably benefit from that is #1317

[mstoykov]

in order for #1525 to work it required to change the config in the Validate function, which shouldn't really change the configuration

[na--]

As mentioned in #1514 (comment) and ee17425, the current re-segmentation has some issues, especially when it comes to the shared-iterations executor.

Fairer re-segmentation is needed, since the current one (based on the striping algorithm) doesn't deal well with small numbers. Running k6-1007 run --vus 2 --iterations 2 --execution-segment 0:1/100 --execution-segment-sequence=0,1/100,1 github.com/loadimpact/k6/samples/http_get.js will result in a 1VU/1iteration test run. So, 1% of the test run would be equal to the other 99%, which is obviously not ideal. We should be safe to leave it as it is for v0.27.0, since it only affects small numbers, but as demonstrated in the commit above, fixing it should be relatively easy and clean once we have the proper abstractions...

[imiric]

Another cleanup reminder (see #1490 (review)): printExecutionDescription() was added to unify the execution description output for both local and cloud execution, but it could use a more integrated approach.