Aggregation on empty groups fails because of LoadMaps

[Daniel Cox]

If I do an Aggregate_SUM on a Group from a DataFrame with no rows, instead of returning 0 like I would expect, it fails at the top of LoadMaps. This is because Aggregation calls LoadMaps at the end to build the resulting DataFrame, but an empty map slice is an error for LoadMaps.

In this case, I think it would be best for Aggregate to never call LoadMaps with an empty map slice, but rather insert the zero value appropriate for whatever AggregationType the user picked for all empty groups.

To reproduce:

import(
	"fmt"
	"github.com/go-gota/gota/dataframe"
	"github.com/go-gota/gota/series"
)

df := dataframe.New(
	series.New([]string{"e", "Pi", "Phi", "Sqrt2", "Ln2"}, series.String, "Strings"),
	series.New([]float64{2.718, 3.142, 1.618, 1.414, 0.693}, series.Float, "Floats"),
)
agg := df.
	Filter(dataframe.F{0, "Strings", "==", "DNE"}).
	GroupBy("Strings").
	Aggregation([]dataframe.AggregationType{dataframe.Aggregation_SUM}, []string{"Floats"})

fmt.Println(agg)

// DataFrame error: load maps: empty array

[Martyn]

just been bitten by this bug, anyone working on a fix?