remove fill_entries! and use LinearAlgebra.fillstored!

[fverdugo]

Now it is a good time to do this before releasing 0.17.

cc @amartinhuertas

[fverdugo]

The only non-obvious part is that the implementation of fill_entries! for SparseMatrixCSR needs to be done in its own package (https://github.com/gridap/SparseMatricesCSR.jl) to avoid type piracy.

[fverdugo]

To think also if we can find Julia equivalents of functions like scale_entries! (i.e. scaling the stored values)

[amartinhuertas]

I will do this

[fverdugo]

For scale_entries! we can use rmul! instead:

help?> rmul!
search: rmul! baremodule parentmodule promote_rule

  rmul!(A::AbstractArray, b::Number)

  Scale an array A by a scalar b overwriting A in-place. Use lmul! to multiply scalar from left. The scaling operation respects the semantics of the multiplication * between an element of
  A and b. In particular, this also applies to multiplication involving non-finite numbers such as NaN and ±Inf.

[amartinhuertas]

See PR #680 for the status of this work

[fverdugo]

Closed via PR #680