[RFC]: add lint rule to ensure `string/format` is provided an expected number of arguments

[kgryte]

In examples,

// Valid:
throw new Error( format( '%s %s', 'foo', 'bar' ) );

// Invalid:
throw new Error( format( '%s %s', 'foo' ) );
throw new Error( format( '%s %s', 'foo', 'bar', 'beep' ) );

In short, for all instances where string/format is invoked, we want the number of trailing arguments to match the number of placeholders in the input string.

When too few or too many arguments are provided, that is likely a user error.

We should skip the scenario where format is invoked via #.call or #.apply, which is very rare in the codebase.

Notes

1. We should be able to use the format-tokenize package to tokenize the first argument, resolve the number of placeholders, and then compare against the number of provided arguments.
2. We should not blanket check against any usage of any format symbol. While that is the common alias for string/format in code, we use format elsewhere; nor should format be considered a reserved word for this purpose.
3. Instead, if a source file (source, test, benchmark, example, etc) contains a string/format import statement, we should (a) resolve the assigned alias, (b) resolve all usages of that variable, taking note that we should guard against shadowing (meaning, it is not enough to simply grep for the assigned alias; you need to find all instances where that particular symbol, not any shadowed symbol, is used), and (c) compare the number of provided args to the number of placeholders. For (b), there have been a few instances where we've needed to resolve symbols in parent scopes which are relevant to a scoped content; we may be able to reuse them, or, better, create a common utility.

[kgryte]

And just to be clear, string/format does not currently require a matching number of arguments, and I am not advocating that it should.

In [13]: format( '%s' )
Out[13]: 'undefined'

Not enforcing a matching number of arguments matches the behavior of printf et al, so I am not of the opinion that updating string/format is the solution here.