update_visit_data: intelligently filter out configured responses?

[shippy]

With the latest influx of variables into Data Entry

• lssaga1_youth_dm15d_dm15d_1_dm15d_y and "lssaga1_youth_dm15d_dm15d_2_dm15d_y have a bunch of In progress responses, some . responses
• mrireport_mri_set*_*_mrilu6_mricns90[dh] have a bunch of . responses.

This is clocking at about 83 issues. The standard SOP is to ask the sites to go fix it in the Import project, but given the homogeneity of the errors, it might not be a bad idea to implement some filtering in update_visit_data. Implementation sketch:

1. a config section in sibis_sys_config.yml of the form

   update_visit_data:
     ignored_values:
       lssaga1_youth_dm15d_dm15d_1_dm15d_y:
         - "."
         - "In progress"

2. a method in update_visit_data that goes through ignored values per field and empties them out with something like data.loc[data[varname] == value, varname] = np.nan