ci: refactor test workflows to use a DRY/matrix approach

[gustavovalverde]

Motivation

We have a bunch of workflow code repeating to execute almost the same tasks, for example the following jobs have over a 100 line of codes, and both are almost a replica:

• test-full-sync
• test-stateful-sync

This also happens with:

• test-all
• test-fake-activation-heights
• test-empty-sync
• test-lightwalletd-integration
• regenerate-stateful-disks

This same behavior is repeating with:
#4223

Designs

• As most test doesn't share a standard cli execution, this makes pipeline standardization a bit harder
• Using entrypoint.sh can workaround most of this issues. And might be the ideal solution until a more standardize way of executing the test is created within zebra/rust/cargo.
• All tests can have an execution variable with an specific nomenclature

Related Work

#4237

[teor2345]

Can this wait until after we've finished the next few lightwalletd integration tests?

• fully synced RPC
• generate lightwalletd cached state
• use lightwalletd cached state

Then we'll know the parameters the matrix or sub-workflow needs.

If we try to do this refactor now, we could end up adjusting parameters a lot, in multiple places.