grid_to_raster should infer the number of cells from the provided meshgrid

[atmorling]

Currently grid_to_raster takes in x and y len params which are used to calculate the raster grid, this can cause unexpected behaviour when combined with create_meshgrid as the calculated raster cells can be different to the meshgrid cells.

ecoscope/ecoscope/io/raster.py

Lines 211 to 222 in cf861ed

	def grid_to_raster(grid=None, val_column="", out_dir="", raster_name=None, xlen=5000, ylen=5000):
	"""
	Save a GeoDataFrame grid to a raster.
	"""
	bounds = grid["geometry"].total_bounds
	nrows = int((bounds[3] - bounds[1]) / ylen)
	ncols = int((bounds[2] - bounds[0]) / xlen)
	if val_column:
	vals = grid[val_column].to_numpy()
	else:
	vals = np.zeros(len(grid))
	vals = np.flip(vals.reshape(nrows, ncols, order="F"), axis=0)

[atmorling]

If we assume the order of the geometry column is always per the output of create_meshgrid ie:

(xmin, ymin)
(xmin, ymin+1)
...
(xmin, ymax)
(xmin+1, ymin)
...
(xmax, ymax)

We could iterate though the geometry and check for the first non-intersection and know that will be the end of our axis, or max_y, then max_x = max_y / len(geometry)

[atmorling]

Per discussion with Jake a while back this approach might be better