element-wise and object-wide subsetting à la `[`

[HelenaLC]

• what's not straightforward (in my mind) is how to deal with multiscales... e.g., if there arrays of dim. 16x16 8x8 and 4x4, only subsets of 1..4, 5..8, 9..12, 13..16 could be propagated across all scales... will have to see Python's peeps thoughts on this
• object-wide subsetting should be implementable, in principle, but isn't straightforward either... every element can have a different dimensionality (e.g., 2 for labels and 3 for images) and dimension (e.g., array extend)...
• one way to implement may be according to these examples...

  x["images",,] # get 'x' with images only
  x[,1,] # get 'x' with first element of each layer only
  x[2,1,list(TRUE, seq(64))] # subset columns of 1st label 
  x[1,c(1, 2), list(list(...), list(...))] # subset images 1-2 differently

[HelenaLC]

1st attempt at implementation in v0.99.14, currently omitting k. I.e., only layers and elements can be subsetted directly, not the contents of any element. E.g.,

x <- file.path("extdata", "blobs.zarr")
x <- system.file(x, package="SpatialData")
x <- readSpatialData(x, tables=FALSE)

# layers are taken in order of appearance
# (images, labels, points, shapes, tables)
x[-4] # drop layer
x[4, -2] # drop element
x["shapes", c(1, 3)] # subset layer
x[c(1, 2), list(1, c(1, 2))] # multiple