The current implementation of data frame metadata using the df.attrs semi-functionality in pandas is useful, but it has a bug/decision that produces an exception when doing dataframe concatenations because some of the metadata is kept as a list. This PR changes the table metadata format to split x/y/z into different fields, similar to how it's actually kept and transmitted. This is a mild breaking change for those who use the voxel_resolution field right now (e.g. dash apps will need a one line fix in a few places), but I think it's useful to avoid a fairly cryptic error.
The current implementation of data frame metadata using the
df.attrssemi-functionality in pandas is useful, but it has a bug/decision that produces an exception when doing dataframe concatenations because some of the metadata is kept as a list. This PR changes the table metadata format to split x/y/z into different fields, similar to how it's actually kept and transmitted. This is a mild breaking change for those who use thevoxel_resolutionfield right now (e.g. dash apps will need a one line fix in a few places), but I think it's useful to avoid a fairly cryptic error.