Add fastPfor decoder to Typescript

[Turtelll]

This PR introduces fastPfor decoding to typesctipt

• testcoverage is reached with unit
• i tested the decoder with a large fastPfor encoded tileset and re-generated the expected testdata and tests pass

[codecov-commenter]

Codecov Report

❌ Patch coverage is 84.36019% with 33 lines in your changes missing coverage. Please review.
✅ Project coverage is 83.80%. Comparing base (d01d7fd) to head (8f449aa).
⚠️ Report is 6 commits behind head on main.

Files with missing lines	Patch %	Lines
ts/src/decoding/fastPforDecoder.ts	91.25%	14 Missing ⚠️
ts/src/decoding/integerDecodingUtils.ts	71.42%	6 Missing and 2 partials ⚠️
ts/src/decoding/integerStreamDecoder.ts	0.00%	6 Missing and 1 partial ⚠️
ts/src/decoding/stringDecoder.ts	0.00%	2 Missing ⚠️
ts/src/decoding/decodingUtils.ts	90.90%	1 Missing ⚠️
ts/src/mltDecoder.ts	0.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #714      +/-   ##
==========================================
+ Coverage   83.35%   83.80%   +0.44%     
==========================================
  Files          57       58       +1     
  Lines        2439     2605     +166     
  Branches      554      588      +34     
==========================================
+ Hits         2033     2183     +150     
- Misses        380      394      +14     
- Partials       26       28       +2     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[mactrem]

Good first draft!

Let's add benchmarks via Benchmark.js to compare the decoding performance with varint encoding based on different datasets

[mactrem]

Add the FastPfor test data as binary files to keep the tests more clean.
Also let's add more test cases to ensure that the implementation is also valid for edge cases.

[Turtelll]

i switched to using binary files and will add more tests

[mactrem]

there are two copies generated (data. slice and intValues) before decoding the actual data. Let's reduce the number of data transformations, because at least we do not need slice. Additionaly, if the offset into the data buffer is aligned (to 4 bytes) no copy is needd at all. One approach to complety avoid the copies is to directly work on the DataView but which leads to a slower data access performance see https://v8.dev/blog/dataview. But first let's add benchmarks (against varint as baseline) and then let's optimize

[Turtelll]

I will start benchmarking the decoder with different tilesets i encoded accordingly. But i'm not sure if we want to have those benchmarks in the repo itself.

[mactrem]

I will start benchmarking the decoder with different tilesets i encoded accordingly. But i'm not sure if we want to have those benchmarks in the repo itself.

We can think about publishing a seperate package for the fastpfor decoder