[FIX] web_kanban: correct count for dataset size

In non grouped mode, a recent fix (fixing race condition in do_search)
introduced a new bug, the dataset size is not correctly set.

This is a quick and dirty fix, I believe that changing the dataset
internals like this commit (and other) do is really not correct.
However, the correct fix (working on the dataset API) is more dangerous
in a stable release.

addons/web_kanban/static/src/js/kanban_view.js

@@ -138,6 +138,7 @@ var KanbanView = View.extend({
                     }));
                 } else {
                     new_ids = _.pluck(data.records, 'id');
+                    self.dataset._length = data.count;
                 }
                 self.dataset.alter_ids(new_ids);
                 self.data = data;
@@ -157,17 +158,20 @@ var KanbanView = View.extend({
 
     load_records: function (offset, dataset) {
         dataset = dataset || this.dataset;        
-        return dataset._model.query(this.fields_keys.concat(['__last_update']))
+        var query = dataset._model.query(this.fields_keys.concat(['__last_update']))
                 .limit(this.limit || false)
-                .offset(offset || 0)
-                .all()
-                .then(function (records) {
+                .offset(offset || 0);
+        return query.all()
+            .then(function (records) {
+                return query.count().then(function(count) {
                     return {
                         records: records,
                         is_empty: !records.length,
                         grouped: false,
+                        count: count,
                     };
-                });
+                })
+            });
     },
 
     load_groups: function (options) {