improvements for heatmap and histogram

[kushalkolar]

heatmap and histogram need some improvements, histogram is currently broken after #78

• heatmaps for data larger than 8192, as discussed before this is a WGPU limitation. Need to stitch together multiple pygfx.Image world objects
• histogram is broken because histogram "data" is different from the graphic data. I think we should subclass DataFeature to make one specific to histograms, and it could handle things like each bin and its values. We could add a method to HistogramGraphic to change the scale_factor for displaying it. We might also need to subclass ColorFeature for histograms so that indexing sets the color of each bin.

[kushalkolar]

Some ideas to help with tiling Image WorldObject for heatmaps:

from math import ceil
from itertools import product

chunk_size = 128

x_chunks = range(ceil(mcorr.shape[1] / chunk_size))
y_chunks = range(ceil(mcorr.shape[2] / chunk_size))

chunks = list(product(x_chunks, y_chunks))
# chunks is the position of each chunk

start_ixs = [list(map(lambda c: c * chunk_size, chunk)) for chunk in chunks]
stop_ixs = [list(map(lambda c: c + chunk_size, chunk)) for chunk in start_ixs]

grid = list()
for start, stop in zip(start_ixs, stop_ixs):
    x0, y0 = start
    x1, y1 = stop
    grid.append(data[:, x0:x1, y0:y1].mean(0))

[kushalkolar]

heatmap being done in #143

[kushalkolar]

So we've decided histogram is not happening because current libs more than good enough for histograms, as far as I've experienced.