Release 1.0 tracking

[bkamins]

In this issue I will track and update things to finish for 1.0 release of DataFrames.jl.

Top priority:

• combine without transformations Disallow combine without any transformations #2633
• finalize joins we need (semijoin and antijoin still) Faster joins meta issue #2340
• improve documentation add Gallery to the documentation #2602 and Rework the manual #2595
• make code more compiler friendly RFC/DNM: profile-guided despecialization #2597 and Split methods to reduce compile cost #2566, and Latency comparison between transform and combine with grouped data frame #2516
• rethink design of stack and unstack Add more keyword arguments to stack and unstack #2422, Stacking multiple groups of columns #2414
• rethink workflows with filtered data frames Improwe workflows with filtered DataFrame #2354 and Skipping missing values more easily #2314
• update precompilation statements (@nalimilan hopefully can help with this), Re-generate precompile statemenst before 1.0 release #2642
• update to Julia 1.7 changes Follow Julia 1.7 broadcasting rules #2643
• tests cleanup fix split-apply-combine tests #2644
• decide if we should use UInt32 instead of Int in GroupedDataFrame and joins when we are sure that there are less groups than UInt32 values. Also consider checking if grouped data frame is sorted before grouping

Nice to have:

• using multi-threading in other operations than split-apply combine Add more threading supoprt #2626
• handling of GroupKey comparisons GroupKey should be comparable between DataFrames #2639
• API of eachcol Usage of eachcol seems to be a pun? #2636
• extend row-like outputs handling in combine et al Discussion regarding custom row-like outputs in combine and transform #2576
• type of array returned in transformations Return array type for combine/transform/select #2569 and Formalize API for column vectors #2567
• grouping PooledArrays improvements Weird slowdowns when grouping on PooledArray #2564
• PrettyTables.jl related houskeeping Post PrettyTables.jl housekeeping #2525
• type instability in aggregation Resolve type instability issues in aggregation #2424
• improve using of aggregation fast path make sure aggregation uses fast path if a single column is used #2423

[bkamins]

I am closing this as now the labeling of 1.0 milestone is precise enough to track it there. Essentially we only need some minor clean-ups before 1.0 release.