dynamic spline core

[j042]

Addressed issues

• Too many types for backend splines in cpp

Description

Instead of making every single classes available, we define a template-free base spline type. Then, we inherit from both base and third party splines. Can dynamically create third party splines, but return, for example, a shared pointer of the base spline. Here, at base spline, we've defined scalar raw-ptr based operations. That way it is compatible with anything.

Frontend changes

til now	from this PR	alias to old version
Spline.elevate_degree	Spline.elevate_degrees	yes
Spline.reduce_degree	Spline.reduce_degrees	yes
Spline.basis_fuctions	Spline.basis_and_support	no
Spline.nearest_pcoords	Spline.proximities	no
Spline.permute_parametric_axes	spline.permute_parametric_axes	no (member function to module function)
splinepy._splinepy	splinepy.splinepy_core	no (new core module name)

Changes

• tracked arrays for spline properties
• c_spline based init
• more bezier operations (need tests)
• core spline info checks
• proximities queries return more info
• proximities for bezier splines
• ND operations form bezier splines
• multithread executions for queries
• nicer logging
• NAME_TO_TYPE
• global vars
• multi patch export using splinelib (need tests)
• formatting

Up next

• CD
• unittests
• gustaf updates
• cmake tests
• explicit template instantiation for SplineLib and BezMan (maybe)

[jzwar]

Although this is a very shallow review (I am sorry), this looks very impressive. Love the tracked array functionality and also how much effort you gave to unify the interfaces <3

Very nice work!!!
LGTM

[j042]

v

[jzwar]

LGTM