Update DEBUG_LOG_FORMAT to correctly zero-pad the microsecond portion

[heisencoder]

Currently, when running dbt with the --debug flag, the microsecond portion of the timestamp does not have leading zeros, which causes jagged formatting. Example:

2020-02-05 22:56:50,953988 (MainThread): Parsing macros/materializations/...
2020-02-05 22:56:50,989795 (MainThread): Parsing macros/materializations/...
2020-02-05 22:56:51,48211 (MainThread): Partial parsing not enabled
2020-02-05 22:56:51,80875 (MainThread): Acquiring new bigquery connection "model.mytable".
2020-02-05 22:56:51,81005 (MainThread): Opening a new connection, currently in state init
2020-02-05 22:56:51,883792 (MainThread): Acquiring new bigquery connection "model.bar".

The fix was to just paste in the text of the TODO immediately above DEBUG_LOG_FORMAT in logger.py.

However, if that wasn't working for some other reason, another fix would be to modify DEBUG_LOG_FORMAT like this:

logger.py:
  DEBUG_LOG_FORMAT = (
    #'{record.time:%Y-%m-%d %H:%M:%S%z}.{record.time.microsecond:06} '

(i.e., change the '03' to a '06' so that 6 digits are zero-padded instead of just 3).

(I also took the liberty to change the European ',' decimal point to an American '.' decimal point.)

[beckjake]

I'm into all this! I've got some scripts that parse dbt logs that I'll have to update, but that's probably worth it. I've kicked off the tests, and when they pass we will merge this for 0.16.0

[heisencoder]

Thanks!

[drewbanin]

@beckjake merge at will :)

Thanks @heisencoder!