Analytically compute the square root of the IWP transition covariance

[nathanaelbosch]

Solves #257

But instead of actually using IntegratedWienerProcesses.jl I copied and slightly changed the code to compute the left square root factor, as well as to automatically use BigInt to compute factorials if the desired smoothness gets too large.

Left to do:

• [x] can we simplify to have less calls to factorial?
• [x] filling by row or column can have quite a performance cost; are we doing it in the right order right now?
• [x] add a docstring

[nathanaelbosch]

Held up by Fenrir.jl's patch release with increased ProbNumDiffEq.jl compat: https://github.com/JuliaRegistries/General/pull/94415

[codecov[bot]]

Codecov Report

Merging #264 (011165b) into main (9e8c8e0) will decrease coverage by 0.19%. The diff coverage is 72.72%.

@@            Coverage Diff             @@
##             main     #264      +/-   ##
==========================================
- Coverage   95.67%   95.49%   -0.19%     
==========================================
  Files          35       35              
  Lines        1435     1444       +9     
==========================================
+ Hits         1373     1379       +6     
- Misses         62       65       +3     

Files	Coverage Δ
src/priors/iwp.jl	95.77% <72.72%> (-4.23%)	:arrow_down:

:mega: We’re building smart automated test selection to slash your CI/CD build times. Learn more