Modifying thusands separator belongs to language settings... So to do it within custom module you need to put desired data in csv, xml or yml file and load that file during module instalation... .

I prefer yml file for this pourpuse....here is an example for croatian language:

-

!python {model: base.language.install}: |
  lang_ids = self.pool.get('res.lang').search(cr, uid, [('code', '=', 'hr_HR')], limit=1, context=context)
  if not lang_ids:
     wizard_id = self.create(cr, uid, {'lang': 'hr_HR'}, context)
     self.lang_install(cr, uid, [wizard_id], context)

-
  I make Croatian as default language if installed
-
  !python {model: res.lang}: |
    if self.search(cr, uid, [('code', '=', 'hr_HR')], limit=1):
      value_obj = self.pool.get('ir.values')
      ids = value_obj.search(cr, uid, [
        ('name', '=', 'lang'), ('key', '=', 'default'), ('model', '=', 'res.partner'), ], limit=1, context=context)
      vals = { 'name': 'lang', 'key': 'default', 'key2': False, 'model': 'res.partner', 'object': False, 'value_unpickle': 'hr_HR'}
      if ids:
        value_obj.write(cr, uid, ids, vals, context)
      else:
        value_obj.create(cr, uid, vals, context)
-
  I correct all formats for Croatian language
-
  !python {model: res.lang}: |
    lang_ids = self.search(cr, uid, [('code', '=', 'hr_HR')], limit=1)
    if lang_ids:
      self.write(cr, uid, lang_ids, {
        'grouping': '[3,0]',
        'decimal_point': ',',
        'thousands_sep': '.',
        'name': 'Hrvatski',
        'date_format': '%d.%m.%Y',
        'time_format': '%H:%M:%S', }, context)

Just copy/paste/modify  in some_file.yml,
and include that file in your data section of __openerp__.py of your custom module...