In https://bugzilla.mozilla.org/show_bug.cgi?id=1499819, it was found
that `ROW_COUNT()` causes db replication to lag. As such, it makes sense
to grep for `ROW_COUNT()` in our stored procedure lint script.
The script was called `explain-warn.js` before, but in light of the new
functionality it's renamed here to the more generic `procedure-lint.js`.

These two scripts use ROW_COUNT() but I couldn't see an obvious way to
work round them. In the case of consumeRecoveryCode_2, ROW_COUNT() is
used as an interim signal for aborting the remainder of the procedure.
In the case of setPrimaryEmail_3, the UPDATE is in a transaction which
causes affectedRows to be zero upon returning.