v2.2.0

Changes since v2.1.3:

• All algorithms now recycle tables using a global table pool. This reduces allocations by ~3-4x in almost all cases, and is a general ~1.3x speedup for raw benchmarks. For general use cases (such as fromString), this can provide an >2x speedup.
• Implements algorithm FastIntegerOutput 1.26, which kicks in at >2200 limbs
• Implements fast path for 2^k modulo
• Many bugfixes regarding Burnikel-Ziegler division
• Fixes .__len off-by-one results

View full changelog here.