Prior to this change we were onbording using a seed phrase, which was not ideal,
as it could lead to a seed phrase leakage in the Playwright report added to the
artifacts of CI jobs (that's why we've temporarily switched off storing of the
test artifacts).
Now that we support in Taho the onboarding with a JSON keystore file (which
stores an encrypted private key), we are good to use this method of account
import in our E2E tests. This method is safer, as the sensitive data entered in
the form during onboarding is obfuscated, so there's no risk it will leak in the
Playwright report (and even if so, what will leak will be just a password, not a
corresponding to it JSON file).
As we're switching to this onboarding method, we can bring back storing of the
Playwrigt reports in the CI jobs artifacts.

Apart from the above, as part of this commit/PR we're also introducing a script
which can be used to generate a JSON keystore file based on the provided private
key and password.

We need to run the install there to avoid complaints from the linter about a
missing module (`Unable to resolve path to module 'ethereumjs-wallet'`).

The transactions e2e tests use the first address of the testing wallet
(`testertesting.eth`). But we already know that in some tests that we plan to
add in the future, we'll want to use a different account - the third address of
the testing wallet (`e2e.testertesting.eth`). So we will need to use names of
the variables and secrets that differenciate those different addresses.

I've pushed by mistake a change commenting out some fragment of the e2e tests
that shouldn't be commented out. I'm fixing it in this commit.

We changed this for production fork builds to differentiate from the release builds, and we updated e2e tests to look for the suffix, but on non release branches the fork builds were not getting the suffix, leading to test failures.