Update: add beforeStatementContinuationChars to semi (fixes #9521)

[mysticatea]

What is the purpose of this pull request? (put an "X" next to item)

[X] Changes an existing rule: fixes #9521.

What changes did you make? (Give an overview)

This PR adds a new option beforeUnreliableLineButSafe to semi rule. This option specifies the behavior about semicolons (or lacking semicolons) which satisfy the following two condition:

1. The semicolon (or lacking semicolon) is followed by the tokens which can make ASI hazard ([, (, +, -, or `).
2. But the semicolon (or lacking semicolon) does not make ASI hazard because the next token does not connect to the previous statement.

For example:

import a from "a"
[1,2,3].forEach(doSomething)

The option can be one of three values:

• { beforeUnreliableLineButSafe: "any" } ignores the semicolons (or lacking semicolon). This is the default. This is the current behavior.
• { beforeUnreliableLineButSafe: "always" } requires the semicolons.
• { beforeUnreliableLineButSafe: "never" } disallow the semicolons.

Is there anything you'd like reviewers to focus on?

The option name beforeUnreliableLineButSafe is not the best name. Does somebody have a better idea?

[not-an-aardvark]

The option name beforeUnreliableLineButSafe is not the best name. Does somebody have a better idea?

Maybe something like beforeStatementContinuationChars? A "statement continuation character" would be something like ( or [. (I'm not sure if there is a standard name for characters like ( and [.)

[not-an-aardvark]

Can you clarify the default value of beforeUnreliableLineButSafe when never is used? (Is it "ignore"? If so, should we add an explicit "ignore" option?)

[mysticatea]

OK.

I was hovering between adding the description of "any" or not because similar omitLastInOneLineBlock option does not have the description about its default value.

[mysticatea]

I updated this PR.

[platinumazure]

LGTM, thanks!