QR decomposition for Least-Squares in LLS sharp

[jpfairbanks]

In the definition of the sharp matrix for LLSDDSharp:
https://github.com/AlgebraicJulia/CombinatorialSpaces.jl/blob/main/src/DiscreteExteriorCalculus.jl#L781

There is a pinv(X'X)

CombinatorialSpaces.jl/src/DiscreteExteriorCalculus.jl

Line 804 in 729946c

LLS = pinv(X'*(X))*(X')

This should be replaced by a QR decomposition of Q. Ideally, we would do an implicit solve with Rx = Q'b but since you want to realize a matrix for this, you could at least take the reduced inv(R)*Q'.