Tighten WoodburyPDMat signature

[dkarrasch]

This came up in a nanosoldier run in JuliaLang/julia#46196. The proposed change should not alter the current behaviour and should be backward compatible, but additionally should make this package "ready" for JuliaLang/julia#46196 in case it gets accepted.

[codecov]

Codecov Report

Merging #81 (838c21a) into main (6d15d47) will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##             main      #81   +/-   ##
=======================================
  Coverage   92.95%   92.95%           
=======================================
  Files          13       13           
  Lines         525      525           
=======================================
  Hits          488      488           
  Misses         37       37           

Impacted Files	Coverage Δ
src/woodbury.jl	100.00% <ø> (ø)

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[sethaxen]

Thanks for the PR! My only concern is whether there are some invocations of qr whose Q is not an AbstractQ. e.g. for some overload for a user-defined matrix. This really isn't an issue for this package, but it is if we upstream this structure to PDMatsExtras as is planned.

The alternative to narrowing the type would be widening, i.e. removing type constraints. What do you think?

[dkarrasch]

Yes, that's another option. I was not aware of any context. Widening is perhaps even the better idea for generic programming.

[sethaxen]

Let's go with widened then. Thanks for the PR!