Handle Implementation Failure Gracefully

Currently, a single failing implementation crashes the whole test harness.

Since some implementations may fail on certain input, we should fix this so that these failures can be reported the summary.

I think the way to fix this is to move all the test cases from the .env file into individual files, and invoke the implementations on test case file individually. We could also consider moving the `.env` file into a `JSON` file of test cases, and then generate the individual .env files from that.