Reduce type flexibility in BsplineManifold

[hyrodium]

This PR fixes #353.

[codecov]

Codecov Report

Attention: 11 lines in your changes are missing coverage. Please review.

Comparison is base (ae5ba38) 93.84% compared to head (d9b7342) 93.79%.

Files	Patch %	Lines
src/_Refinement.jl	72.22%	5 Missing ⚠️
src/_RationalBSplineManifold.jl	71.42%	4 Missing ⚠️
src/_BSplineManifold.jl	81.81%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #355      +/-   ##
==========================================
- Coverage   93.84%   93.79%   -0.06%     
==========================================
  Files          14       14              
  Lines        1642     1660      +18     
==========================================
+ Hits         1541     1557      +16     
- Misses        101      103       +2     

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

[hyrodium]

Before this PR

julia> using BasicBSpline

julia> P1 = BSplineSpace{2, Int}(knotvector"3 113")
BSplineSpace{2, Int64, KnotVector{Int64}}(KnotVector([1, 1, 1, 3, 4, 5, 5, 5]))

julia> P2 = BSplineSpace{2, Rational{Int}}(knotvector"3 113")
BSplineSpace{2, Rational{Int64}, KnotVector{Rational{Int64}}}(KnotVector(Rational{Int64}[1//1, 1//1, 1//1, 3//1, 4//1, 5//1, 5//1, 5//1]))

julia> a = rand(dim(P1), dim(P2))
5×5 Matrix{Float64}:
 0.741979   0.263108  0.316128   0.372952  0.421981
 0.486815   0.67646   0.328703   0.451822  0.340635
 0.0072873  0.962419  0.128144   0.163797  0.621682
 0.605695   0.600226  0.0666295  0.745873  0.0474121
 0.546628   0.018101  0.305919   0.290262  0.884664

julia> bsplinespaces(BSplineManifold(a, P1, P2))[1]
BSplineSpace{2, Int64, KnotVector{Int64}}(KnotVector([1, 1, 1, 3, 4, 5, 5, 5]))

julia> bsplinespaces(BSplineManifold(a, P1, P2))[2]
BSplineSpace{2, Rational{Int64}, KnotVector{Rational{Int64}}}(KnotVector(Rational{Int64}[1//1, 1//1, 1//1, 3//1, 4//1, 5//1, 5//1, 5//1]))

After this PR

julia> using BasicBSpline

julia> P1 = BSplineSpace{2, Int}(knotvector"3 113")
BSplineSpace{2, Int64, KnotVector{Int64}}(KnotVector([1, 1, 1, 3, 4, 5, 5, 5]))

julia> P2 = BSplineSpace{2, Rational{Int}}(knotvector"3 113")
BSplineSpace{2, Rational{Int64}, KnotVector{Rational{Int64}}}(KnotVector(Rational{Int64}[1//1, 1//1, 1//1, 3//1, 4//1, 5//1, 5//1, 5//1]))

julia> a = rand(dim(P1), dim(P2))
5×5 Matrix{Float64}:
 0.933547    0.324843  0.977758   0.182171  0.55386
 0.404299    0.911178  0.0584839  0.615514  0.264623
 0.00691595  0.896858  0.102844   0.553428  0.226851
 0.846493    0.987066  0.476708   0.99308   0.129281
 0.301458    0.242925  0.420328   0.224337  0.957864

julia> bsplinespaces(BSplineManifold(a, P1, P2))[1]
BSplineSpace{2, Rational{Int64}, KnotVector{Rational{Int64}}}(KnotVector(Rational{Int64}[1//1, 1//1, 1//1, 3//1, 4//1, 5//1, 5//1, 5//1]))

julia> bsplinespaces(BSplineManifold(a, P1, P2))[2]
BSplineSpace{2, Rational{Int64}, KnotVector{Rational{Int64}}}(KnotVector(Rational{Int64}[1//1, 1//1, 1//1, 3//1, 4//1, 5//1, 5//1, 5//1]))

[hyrodium]

Sometimes type promotions may fail because

julia> AbstractKnotVector{Int}(UniformKnotVector(1:5))
ERROR: MethodError: no method matching AbstractKnotVector{Int64}(::UniformKnotVector{Int64, UnitRange{Int64}})

Closest candidates are:
  AbstractKnotVector{S}(::KnotVector{T}) where {S, T}
   @ BasicBSpline ~/.julia/dev/BasicBSpline/src/_KnotVector.jl:93

Stacktrace:
 [1] top-level scope
   @ REPL[9]:1

One can just modify the arguments manually not to require type promotions (_promote_knottype).

[hyrodium]

RationalBSplineManifold was also updated.