[FEATURE] Add half_float sort queries in benchmark.

[gashutos]

Is your feature request related to a problem?

half_float field type is unique and we have custom comparators for this field. half_float is a mapping where BKD is indexed with 2 bytes while its doc values are stored as 4 bytes, so the sort logic comparison is divided in 2 ways, but still we were able to enable BKD point based optimization for this type.

What solution would you like?

nyc_taxis workload has many float type, tip_amount generally are between 0 to 100 $ and it should qualify as half_float. So making this field as half_float and adding asc & desc sort queries for this field will likely to improve our benchmarking coverage.

Do you have any additional context?

Add any other context or screenshots about the feature request here.