Revert our changes and use FireMasterK's improvements

[Stypox]

This PR does these things:

• it reverts all of our commits related to parsing custom JSON format. You can see all of our commits from here:
  • | 72c8ed4 Revert "Support JS-like JSON"
  • | 41f29d8 Revert "Support keys starting with a non-ASCII character"
  • | d3b1b70 Revert "Support single quote strings"
  • | f9d8d68 Revert "Support keywords as keys"
  • | e2cd3f0 Revert "Fix some more semi-string parsing bugs"
  • | 714bb89 Revert "Disallow semi-strings as value"
  • | b52e21a Revert "Fix Checkstyle errors"
  • | 9718d33 Revert "Fix issue for semi-strings as well"
• the only two commits that were not reverted are 0e64c2b d5da581, which are both also present in FireMasterK's fork
• additionally, the README changes were retained (5494f3c) and updated with fcb2efd
• pulls in updates from upstream
• pulls in @FireMasterK's performance changes (note that this required reverting the other commits, since @FireMasterK's fork has just removed those commits directly, but I would prefer not to force-push on our master branch).
• I skipped commit c624b79 since it pulls in the fastutil library to squeeze out the last bits of performance, but we'd prefer not to increase APK size, as suggested by @FireMasterK in Use Piped's fork of nanojson NewPipeExtractor#981 (comment), also see TeamNewPipe/NewPipeExtractor@325c07d (#981)
• I did include commit 24eb336 though, even though it adds the fastdoubleparser library. I don't think it matters much whether we keep it or not, the JAR is just 191kB but I'm pretty sure most of it would be optimized out or merged during build. Let me know if I should revert this.
• I also skipped commit a507525 and instead made sure we retain compatibility with JDK 11 with 2a27a17.
• fixes checkstyle

All tests pass. This PR is basically a replacement for TeamNewPipe/NewPipeExtractor#981, where many people suggested updating our own fork. While I initially was against, I now think it's necessary if we want to have some of @FireMasterK's commits but not all.

Closes #2

[FireMasterK]

Nice, however in NewPipeExtractor you will need to add support for LazyStrings, like this: FireMasterK/NewPipeExtractor@165b459

[TobiGr]

What is the correct review process for this PR?
I think that the changes which are explained in the description sound reasonable. Reviewing the code line by line isn't necessary in my opinion because it has already been reviewed and is running in Piped. I think this is good to go if the tests in /src/test pass.

I did include commit 24eb336 though, even though it adds the fastdoubleparser library. I don't think it matters much whether we keep it or not, the JAR is just 191kB but I'm pretty sure most of it would be optimized out or merged during build. Let me know if I should revert this.

Including this is ok for me. If you want, feel free to compile two JARs and compare their sizes. I'd also assume that most of the lib is going to be optimized away.

[Stypox]

@TobiGr the difference seems to be just 664 bytes which is totally fine in my opinion:

34626  nanojson-1.11-SNAPSHOT_with_fastdoubleparser.jar
33962  nanojson-1.11-SNAPSHOT_without_fastdoubleparser.jar

[TobiGr]

Tests pass locally and the commits were chosen reasonably.

[Stypox]

I removed the commit that updates to Java 17, and pushed another commit for the pom.xml that makes sure that, independently of the JDK used during build, the resulting JAR is compatible with Java 11. There was only one small thing in the whole codebase that was incompatible with Java 11, so I fixed that.