Grammar - fix MID statement

[mansellan]

Closes #4128

There were a few issues:

1. Was using ASSIGN instead of EQ
2. Was using lExpression instead of expression
3. Was not referenced from mainBlockStmt
4. No tests

[bclothier]

IMO, we need negative tests, too to assert that an expression that contains a Mid() function does not get parsed as a mid statement. We also need equivalent tests for Mid$() statement/function.

For example, we definitely don't want something like If Mid$("foo", 1) = "f" Then to be treated as a Mid statement. The tests need to validate this.

[codecov]

Codecov Report

Merging #4129 into next will increase coverage by <.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##             next    #4129      +/-   ##
==========================================
+ Coverage   52.55%   52.55%   +<.01%     
==========================================
  Files         963      963              
  Lines       33142    33140       -2     
==========================================
- Hits        17415    17414       -1     
+ Misses      15727    15726       -1

Impacted Files	Coverage Δ
...rsing/Binding/Bindings/SimpleNameDefaultBinding.cs	85.25% <100%> (+0.38%)	⬆️
...ar/PartialExtensions/VBAParserPartialExtensions.cs	53.3% <0%> (ø)	⬆️

[bclothier]

Also, make sure to check whether you can remove the "hack" from the resolver

Rubberduck/Rubberduck.Parsing/Binding/Bindings/SimpleNameDefaultBinding.cs

Lines 78 to 85 in c46cd4d

	//TODO - this is a complete and total hack to prevent `Mid` and `Mid$` from creating undeclared variables
	//pending an actual fix to the grammar. See #2618
	else if (!_name.Equals("Mid") && !_name.Equals("Mid$"))
	{
	var undeclaredLocal = _declarationFinder.OnUndeclaredVariable(_parent, _name, _context);
	return new SimpleNameExpression(undeclaredLocal, ExpressionClassification.Variable, _context);
	}
	return new ResolutionFailedExpression();

[mansellan]

Guessing that (function:Variant) is due to function aliasing. But it's not resolving it to a variable now :-)

[bclothier]

I wonder if it's necessary to have additional unit tests to prove that it's not an undeclared variable to directly test the effect of the hack that was removed?

[retailcoder]

Approving because resolving as a function is a huge step forward compared to resolving to an undeclared variable (i.e. not resolving it at all), however Mid$ needs to resolve to a String-returning function; the inspection for string-typed functions will be having false negatives until that's fixed.