There are two flavours of `match?`. If the receiver of `match?` has type String
then the argument to `match?` is a regular expression. However, if the receiver of
`match?` has type Regexp then the argument is the text.
The role of receiver and argument flips depending on the type of the receiver, this
caused a lot of false positives when looking for string-like literals that are
used as a regular expression.
This commit attempts to improve things by trying to determine whether the type of the
receiver is known to be of type Regexp. In such cases we know that the argument
is unlikely to be regular expression.
