Add `get_column_data_column()` and `set_column_data_column()` to mimic `se$`

[LTLA]

The R package allows people to get/set the column data with:

se$X <- runif(ncol(se))
se$X

as a shorthand for colData(se)$X. Doing the same for Python would allow users to replace

se.get_column_data().get_column("X")
se.set_column_data(se.get_column_data().set_column("X", ...)))

with just

se.get_column_data_column("X")
se.set_column_data_column("X", ...)

The second one is particularly convenient.

[LTLA]

Or support an optional argument:

se.get_column_data("X")
se.set_column_data("X", <column>)

Latter just checks if first arg is a string to decide whether to replace the entirety of the coldata or just a single column therein.

Could also extend this to integers to quickly get/set by position.

Also do the same for the row data.