Non uniformely spaced interpolation grid

[terliuk]

**Describe **
In current implementation of interpolation maps with grids, we allow setting either list of points or np.linspace arguments

straxen/straxen/itp_map.py

Lines 121 to 129 in 25e70e3

	elif isinstance(csys[0][0], str):
	# Support for specifying coordinate system as a gridspec
	grid = [np.linspace(left, right, points)
	for _, (left, right, points) in csys]
	csys = np.array(np.meshgrid(*grid, indexing='ij'))
	axes = np.roll(np.arange(len(grid) + 1), -1)
	csys = np.transpose(csys, axes)
	csys = np.array(csys).reshape((-1, len(grid)))
	self.dimensions = len(grid)

,
this prevents us from giving non-uniform binning, or mix of uniform and not uniform.

For example, something like :

{...
    'coordinate_system' : [ 
          ["a", [0.0, 10.0, 101] ] , 
          ["b", [0.0, 10., 20., 30.,40.,41, 42, 43, 44, 45, 46, 47,48,49, 50] ] 
     ],  
...}

where for b i give interpolation centers. Such way, we can make a variably spaced interpolation, what can help to reduce size of maps. For example, when there is a predominantly smooth parameter with a rapidly changing values in limited regions (R, Z, energy etc) .

[dachengx]

Hey @terliuk. Nice idea! To avoid misunderstandings. Would you please provide a map with your mentioned coordinate_system for testing? Thanks!