perly - simplify non-terminal bare_statement_for

[happy-barney]

By extracting various declarations of cursor variable(s) into dedicated non-terminal
for statement now contains only 3 branches:

• for (;;)
• for ()
• for cursor ()

with LALR(1) parser cursor cannot be empty -> it will produce conflict with for (;;) form.

• This set of changes does not require a perldelta entry.

[bram-perl]

Commit [parser] bare_statement_for - move remember right after keyword forming sequence KW_FOR remember.

This commit message is still a bit too cryptic for me.

Non-terminal remember is empty non-terminal with action starting
new full lexical scope. Such empty non-terminals, when used in
multiple branches, will generate shift/reduce conflict when not
appearing after distinguishable sequence of terminals.

I simply can't understand most of that..
Reading the Such empty non-terminals, when used in multiple branches, will generate shift/reduce conflict makes me wonder if there is/was an issue with how it was written before and/or if there is any change to how the construct is now treated.

Change formalizes KW_FOR as a keyword starting new lexical scope.

The commit message should also what the consequence of this is. (Does it change anything? If yes then what? If no then explicitly mention it in the commit message).

Also missing a bit of the "why"; What was the real problem (if any) with the old code?

[bram-perl]

Commit [parser] clause_mexpr - deduplicate shared pattern into non-terminal PERLY_PAREN_OPEN mexpr PERLY_PAREN_CLOSE

For this commit I would use something like:

Add `clause_mexpr` to remove duplicated code

The pattern `PERLY_PAREN_OPEN mexpr `PERLY_PAREN_CLOSE` is repeated in several places
add `clause_mexpr` to avoid this repetition.

This change should not affect any of the behaviour. 

In my opinion having: [parser] clause_mexpr and non-terminal and PERLY_PAREN_OPEN mexpr PERLY_PAREN_CLOSE in the commit title does not add any real value. (The title is too long and reading it in git log --oneline I would really need to think hard what it really means and if it might alter behavior or not)

[bram-perl]

Commit [parser] bare_statement_for - rename variable(s) product to cursor to visualize possible deduplication

To me (but I'm unfamiliar with the parser) variable(s) product is meaningless/confusing.

I also find the word cursor confusing (but again could be because I'm unfamiliar with the code and/or non-native English). I don't know what to expect when I see the word cursor..

[bram-perl]

Commit [parser] clause_for_cursor - encapsulate FOR loop cursor variables

This is already an improvement;
What I would still add in the commit message is what branches are now merged.

I.e. something like:

Merge the for-branches of:
- `for $scalar (...)`
- `for my $scalar (...)`
- `for my ($foo, $bar) (...)`
- `for \ $scalar (...)`
- `for my \ $scalar (..)`

into a single branch with the parsing of the variables handled  by `clause_for_cusror`