Add support for pd.Grouper() in DataFrame.groupby()

[sethmlarson]

• Alias eland.Grouper() to be the same as pd.Grouper() for convenience
• Eventually will be used for Grouper(..., freq="...") with timestamp bucketing.

[axelv]

Would be interesting if pd.Grouper(key="@timestamp", freq="year") could translate to a date_histogram source in the composite aggregation together with other term sources!