Revert "Permit more control characters in comments (toml-lang#924)"

This reverts commit ab74958.

I'm a simple guy. Someone reports a problem, I fix it. No one reports a problem? There is nothing to fix so I go drink beer.

No one really reported this as a problem, so there isn't anything to fix. But it *does* introduce entirely needless churn for all TOML implementations. Do we need to forbid *anything* in comments? Probably not. In strings we probably only need to forbid \x00. But at least before it was consistent with strings, and more importantly, what everyone wrote code for, which is tested, and already works.

And [none of the hypotheticals](toml-lang#567 (comment)) on why this is "needed" are practical issues people reported, and most aren't even fixed: a comment can still invalidate the file, you must still parse each character in a comment as some are still forbidden, the performance benefits are very close to zero they might as well be zero, and you still can't "dump whatever you like" in comments.

So it doesn't *actually* change anything, it just changes "disallow this set of control characters" to ... another (smaller) set. That's not really a substantial change. The only (minor) real-world issue that was reported (from the person doing the Java implementation) was that "it's substantially more complicated to parse out control characters in comments and raise an error, and this kind of strictness provides no real advantage to users". And that's not addressed at all with this.

---

And while I'm at it, let me have a complaint about how this was merged:

1. Two people, both of whom actually maintain implementations, say they don't like this change.
2. This is basically ignored.
3. Three people continue written a fairly large number of extensive comments, so anyone who wasn't already interested in this change unsubscribes and/or goes 🤷
4. "Consensus".

Sometimes I feel TOML attracts people who like to argue things from a mile-high ivory tower with abstract arguments that have only superficial bearing to actual pragmatic reality.

Fixes toml-lang#995

CHANGELOG.md

@@ -3,7 +3,6 @@
 ## unreleased
 
 - Clarify Unicode and UTF-8 references.
-- Relax comment parsing; most control characters are again permitted.
 - Allow newline after key/values in inline tables.
 - Allow trailing comma in inline tables.
 - Clarify where and how dotted keys define tables.

toml.abnf

@@ -37,10 +37,11 @@ newline =/ %x0D.0A  ; CRLF
 
 ;; Comment
 
-comment = comment-start-symbol *allowed-comment-char
 comment-start-symbol = %x23 ; #
-allowed-comment-char = %x01-09 / %x0E-7F / non-ascii
 non-ascii = %x80-D7FF / %xE000-10FFFF
+non-eol = %x09 / %x20-7E / non-ascii
+
+comment = comment-start-symbol *non-eol
 
 ;; Key-Value pairs
 

toml.md

@@ -57,9 +57,8 @@ key = "value"  # This is a comment at the end of a line
 another = "# This is not a comment"
 ```
 
-Comments may contain any Unicode code points except the following control codes
-that could cause problems during editing or processing: U+0000, and U+000A to
-U+000D.
+Control characters other than tab (U+0000 to U+0008, U+000A to U+001F, U+007F)
+are not permitted in comments.
 
 ## Key/Value Pair