Streamline and update routines in hourly utils

[rjhd2]

During development for #79, found that the remove_missing_data_rows() wasn't doing anything. "Null", "null" "nan" and NaN are all used throughout these scripts without much consistency - and so some search/replace actions don't work. Also some fillna() actions work on the entire dataframe which has knockon impacts downstream.

Finally, many actions could have an inplace=True option set, which might speed up and reduce memory requirements.

With some known good outputs, go through to tidy and tweak.

[sjnoone]

Sorry I can't remember why I wrote that way. we can reveiw after r.06 is complete