Define apply function that does not modify arguments

[apkille]

In preparation for the Gaussian quantum information package. apply would be useful to define here, as other packages in QuantumSavory use apply!.

[apkille]

@Krastanov pinging for review.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 17.29%. Comparing base (521a05b) to head (4cda081).
Report is 1 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main      #30   +/-   ##
=======================================
  Coverage   17.29%   17.29%           
=======================================
  Files          12       12           
  Lines         399      399           
=======================================
  Hits           69       69           
  Misses        330      330           

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

[Krastanov]

I think this should just be *. That is the equivalent used in QuantumOptics and QuantumClifford.
Also, if it just happens that you have a structure that is not capable of in-place modifications, it is still fine to use a method with a ! in the name (for the sake of having a small API). In general, the safe way to use an in-place modifying function would be to write something like

my_structure = my_function!(my_structure)

Sure, this is redundant for the cases where my_structure is mutable, but it is pretty useful when down the line you need to introduce a structure that is not mutable without introducing a new API.

Regrettably, this is just a convention -- there is no way to enforce it in this language yet.

Thus, my preference would be to close this PR

Apologies for the slow response.

[apkille]

@Krastanov You make fair points. My only reasoning for introducing this into the API was for completeness. There's probably exceptions to this rule, but IMO if you have an in-place function you should usually have a corresponding out-of-place function. Since it's already implemented in QO and QuantumClifford, I agree with just using *.

[Krastanov]

yup, good point. As a precedent, * and mul! from LinearAlgebra follow the same separation