Fix closure argument list parsing bug in preferForLoop rule

[calda]

This PR fixes #1709.

This code was being parsed incorrectly:

dict.forEach { (header: (key: String, value: String)) in
    print(header.key)
    print(header.value)
}

so was unexpectedly converted to:

for (header, value) in dict {
    print(header.key)
    print(header.value)
}

instead of:

for header in dict {
    print(header.key)
    print(header.value)
}

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 95.20%. Comparing base (f487996) to head (adea102).

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #1712      +/-   ##
===========================================
+ Coverage    95.13%   95.20%   +0.07%     
===========================================
  Files           20       20              
  Lines        23124    23135      +11     
===========================================
+ Hits         21998    22025      +27     
+ Misses        1126     1110      -16     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[nicklockwood]

I think maybe this only works by coincidence? The key/value labels happen to match the defaults in Dictionary, but what if different labels had been used instead?

[calda]

Just checked -- if you reassign the labels, the compiler emits a warning suggesting you not do this:

let dict = ["foo": "bar"]

// ⚠️ warning: tuple conversion from '(key: String, value: String)' to '(a: String, b: String)' mismatches labels
dict.forEach { (header: (a: String, b: String)) in
    print(header.a)
    print(header.b)
}

So while renaming tuple keys in that way is technically possible, it seems like an anti-pattern that I think is ok for us to not support.

[alistra]

LGTM