Update ec network normalize to support variable subsitutions

[Nospamas]

Updated crmprtd based on bc_hydro network variable substitutions across to the ec network as well.

Current setup does the following variables:

'air_temperature_yesterday_high': 'air_temperature'
'air_temperature_yesterday_low': 'air_temperature'
'total_precipitation': 'total_precipitation'
'wind_direction': 'wind_from_direction'
'wind_gust_speed': 'wind_speed'

Some unknowns for me due to unfamiliarity with the project leaves me with the following questions:

• Will the duplicated air_temperature be an issue? Will cell method be correctly pulled?
• A number of the variables coming back from EC in the XML sheet should maybe be 0 not empty, specifically the precipitation related variables. This could do with a sanity check I think
• I've included total_precipitation in substitutions for completeness but it doesn't switch it to anything
• There remains a bug? with one of the tests (test_download_cache) related to timezones, it fails on my machine due to not being in UTC I think.

[jameshiebert]

Looks great overall, but just a (hopefully) small change to request.

[jameshiebert]

I'm always skeptical when a patch adds LOCs to our repo. :) The approach you've taken here is reasonable, but I don't think that this is (or will be) specific to ECCC. Could you take what you've done here and incorporate it into the swob_ml module? Then if other networks change their variable names in the future, we will have an easy place to add variable mappings.