Core/ircparser: Only trim whitespace from the end of the last parameter.

[genius3000]

In Brief

• Current behaviour is trimming both leading and trailing spaces.
• Corrected behaviour only trims trailing whitespace, as was the intention going by the commit messages and comment.

Impact

Criteria	Rank	Reason
Impact	★☆☆ 1/3	Prevents unwanted trimming of leading whitespace when trailing whitespace existed.
Risk	★☆☆ 1/3	Can't think of anything.
Intrusiveness	★☆☆ 1/3	Small change; diffs look like it won't affect other PRs even in the same file.

Rationale

Essentially the same as the In Brief section. Trimming trailing space is rarely, if any, noticed but trimming leading spaces was unintentional and can ruin a line's formatting.
Reference Issue #1733.

Testing

Had a MOTD line with a trailing space that provided a nice test case:

2021-06-26 05:18:49 [Debug] orig: "     ,'__/_ ___ _  `. "
2021-06-26 05:18:49 [Debug] orig trim:  ",'__/_ ___ _  `."
2021-06-26 05:18:49 [Debug] new trim:  "     ,'__/_ ___ _  `."

Disclaimer

I know this change is kind of ugly, with a trim function implemented directly here. I did not see anywhere else that would benefit from a new Quassel function of rtrim. Totally open to other ideas though.

[deviant]

Oh, I think this will fix a bug I noticed where Quassel would for some reason strip the indentation from a few lines of the MOTD of a server I'm on. I understand why, now: these lines had trailing whitespace, which triggered the stripping logic.

[justjanne]

Why is it even trimmed at all? Ideally we'd keep the content completely intact.

[genius3000]

@justjanne from the commits, it was initially due to trailing spaces in a NAMREPLY causing empty nicks to be created.
Commits 9fe8b83 and 9e637b2.
Those are from a number of years ago and this trimming is just a safety from edge cases (imo). Is this something we can build a test against (I'm completely unfamiliar with the/any test suite) to be fairly certain we don't introduce any regressions by removing this trimming?