Method to aggregate SFE object with spatial grid

[lambdamoses]

The sf package can make rectangular or hexagonal grid. For smFISH data, if we have transcript spot coordinates, they should be used for the spatial binning because often transcripts are not uniformly distributed within the cell. If transcript spots are unavailable, then what to do with cells that only partially overlap with the bin? Parallel processing would help though I kind of feel like BiocParallel isn't so efficient and often slower than good old SerialParam on my laptop.

Other geometries such as cell polygons can also be used when we have subcellular data like from Stereo-seq or Visium HD, although in those cases, as the original data is in a regular rectangular grid, it's better represented as raster and the terra package might be more efficient in aggregating data into cell level.

Closely related to JEFworks-Lab/SEraster#3, so might simply suggest SEraster after it gets to Bioconductor.