Refactor LinkParser

[Nerixyz]

Description

This PR is the continuation of #4507 and #4436.

These are the main changes:

• Use custom parsing for IPv4 addresses over a regex.
• Drop IPv6 support (was disabled by default).
• Use QStringView over QStringRef
• Removed usage of goto.

In my tests (chatterino-benchmark --benchmark_filter=LinkParsing --benchmark_repetitions=100), this version is about 6x faster:

Before

Benchmark	Time	CPU	Iterations
BM_LinkParsing	33159 ns	33692 ns	21333
...
BM_LinkParsing_mean	33569 ns	33523 ns	100
BM_LinkParsing_median	33178 ns	32959 ns	100
BM_LinkParsing_stddev	1520 ns	1457 ns	100
BM_LinkParsing_cv	4.53 %	4.35 %	100

After

Benchmark	Time	CPU	Iterations
BM_LinkParsing	5137 ns	5022 ns	112000
...
BM_LinkParsing_mean	5369 ns	5352 ns	100
BM_LinkParsing_median	5324 ns	5301 ns	100
BM_LinkParsing_stddev	273 ns	254 ns	100
BM_LinkParsing_cv	5.08 %	4.75 %	100

As you can see, I didn't change the isValidTld it's now the bottleneck here. The main reason is that a QString needs to be created and case-converted. Ideally, it could use a QStringView and have case-insensitive search functions.

[github-actions]

clang-tidy made some suggestions

[pajlada]

QChar::isDigit implementation seems to handle unicode stuff, your solution seems good to me

[Nerixyz]

Right, isDigit also checks if the character is the "Nd" unicode category.

[pajlada]

Code looks good, some small nitpicks that you can change if you feel like it's reasonable

[github-actions]

clang-tidy made some suggestions