We should fail on lint errors, then we can follow up in another commit
fixing these errors as part of a PR to enforce the rules.

This commit is large since eslit --fix was run to auto fix most issues.
Some that did not have auto fixes were done manually.

Some changes were not made as they were too risky to change so instead
some in-line comments were used to suppress them. This was preferred
over globally turning off the rule, as keeping it on means new code has
to follow the rule.

Ran yarn prettier --write src test __test__