fixed an inconsistency in the template counts for the usage dashboard

[xsrust]

After some digging, it turned out the inconsistency was twofold.

First, when generating the statistics objects, we organise the templates by their family_id but we save the stats objects by template.title. This means when both customized and non-customized versions of a template are used within an organisation, we save two separate counts, with the same name, ex:

{"count"=>8,
  "date"=>"2019-05-31",
  "by_template"=>
   [{"name"=>"BBSRC Template", "count"=>1},
    {"name"=>"EPSRC Data Management Plan", "count"=>1},
    {"name"=>"University of Manchester Generic Template", "count"=>3},
    {"name"=>"Wellcome Trust Template", "count"=>1},
    {"name"=>"Wellcome Trust Template", "count"=>1},
    {"name"=>"TEST New funder template", "count"=>1}]},

Then, when we're re-martialing the data for the plans_by_template chart, we iterate, identifying the different templates by name, and thus over-write the data with the last count.

It might be worthwile to adjust the way we save the statistics to differentiate between raw and customized funder templates, but for the meantime, I adjusted the template-chart helper to just add the counts when it encounters duplicate entries.

[xsrust]

Addresses #2083

[briri]

great catch! Thanks for digging into this @xsrust