Add compress() function to decrease the file size of BinaryCIF files

[padix-key]

When data is set in BinaryCIFFile objects (e.g via set_structure()), a simple default encoding is used which does not compress the data. Although this is fast, the resulting files can be significantly larger. This PR introduces the compress() function, which takes an existing BinaryCIF... object and creates BinaryCIF... object with size-optimized encoding.

The resulting files are significantly smaller than the uncompressed ones and even quite a bit smaller than the ones originating from the PDB. The following example shows the size for 1L2Y:

Original file from PDB:   199893 bytes
Uncompressed file:       1299713 bytes
File after `compress()`:  169448 bytes

[codspeed-hq]

CodSpeed Performance Report

Merging #674 will not alter performance

_{Comparing padix-key:compress (8320714) with main (231eefe)}

Summary

✅ 44 untouched benchmarks

🆕 1 new benchmarks

Benchmarks breakdown

	Benchmark	main	padix-key:compress	Change
🆕	benchmark_compress	N/A	724.2 ms	N/A