fix(#984): speed up incorrect-bytes-format lint

[ext11n7]

What this PR does

This PR speeds up the incorrect-bytes-format lint.

The main optimization is in incorrect-bytes-format.xsl:

• narrows the selection to leaf o nodes only
• removes unnecessary string-join(...) calls
• replaces a more expensive regex-based validation with a cheaper structural check for byte tokens

Why

The lint was spending too much time scanning nodes and validating byte strings in a costly way during benchmark runs.
This change keeps the behavior the same while reducing the amount of work done per matched object.

Validation

• relevant functional tests passed
• benchmark was re-run after the change
• incorrect-bytes-format became noticeably faster than before

[ext11n7]

@yegor256

[yegor256]

@volodya-lombrozo take a look please

[0crat]

@ext11n7 It is not a good idea to name Git branches the way you named this one: "incorrect-bytes-format". You've earned -6 points. Next time, better give your branch the same name as the number of the ticket that you are solving. In this case, a perfect name, for example, would be "1016". Your running score is +10; don't forget to check your Zerocracy account too).