[python] builder display_summary(): include spatial collections

@jahilton noticed that the builder's summary pivot table didn't include spatial datasets, making it incomplete and also discrepant with the total_cell_count shown in the same report (which did count spatial). This diff includes spatial in the pivot table, breaking it out by collection. Revised example summary: ``` label value census_schema_version 2.4.0 census_build_date 2025-11-04 dataset_schema_version 7.0.0 total_cell_count 217768036 unique_cell_count 125463259 unique count attribute assays cell types cells datasets tissues organism collection callithrix_jacchus census_data 1 40 2275451 27 33 homo_sapiens census_data 37 898 158982719 1086 417 census_spatial_sequencing 2 228 3042411 361 49 macaca_mulatta census_data 2 54 7010229 19 29 mus_musculus census_data 16 473 43653561 172 101 census_spatial_sequencing 2 67 2645566 179 6 pan_troglodytes census_data 1 25 158099 1 1 ``` (2275451+158982719+3042411+7010229+43653561+2645566+158099=217768036) AFAIK this particular report was only used internally, but it was good to explore this because it does raise the question of whether the `census_obj​​["census_info"]["summary"]`, `census_obj["census_info"]["summary_cell_counts"]`, and other `census_info` tables should similarly count the spatial and non-spatial collections separately for each organism, instead of lumping them together as they do now. A separate issue we can file @brianraymor