I looked at the failures in the checker, scanner and emitter. All the checker ones can be trivially split into two expressions -- they are all of the form if (!x.id) x.id = counter++;

The ones in the scanner all look like this:

return pos++, token = SyntaxKind.PercentToken;

(there are a lot of the form return pos += 2, ... as well that should be changed -- see below about a no-comma rule)
Changing this would be a improvement. I think a valid transformation is

pos++;
return token = SyntaxKind.PercentToken;

But that's based on my understanding of , as equivalent to begin/progn/>>. In other words, the pos++ is gratuitously nested inside an expression for no good reason except to save vertical space. (Note that if we had a no-comma rule already, then the scanner wouldn't have any hits. Adding a no-comma rule would also be an improvement in my opinion.)

The emitter uses prefix ++ quite a bit (though by no means exclusively). I suppose we could argue about which one to allow if ++i outnumbers i++.

👍

TS shouldn't get in the way of what is completely traditionally valid JS/Java/C code.
I use ++ & -- operators in many places for casual coding.

This is for our own codebase, not for users.

Whew! Sorry! O_o
Although I think contributors shouldn't have so many rules to follow and have some reasonable freedom.

This is superceeded by #6212 now.