use grep to filter out pointer events

[ces42]

On my computer moving the cursor with the touchpad or scrolling would generated quite high (~10%) cpu usage from fusuma. It seems like this is due to the large amount of output generated by libinput debug-events. Removing all of the POINTER_MOTION and POINTER_SCROLL_FINGER events with grep fixed that for me (I guess grep is much more cpu-efficient than ruby...).

I already wrote this code for use on my computer so I wasn't sure it was necessary to open an issue before creating this PR

• Follow the instructions in CONTRIBUTING. Most importantly, ensure the tests and linter pass by running bundle exec rspec and bundle exec rubocop.
  This doesn't introduce any rubocop complaints but rspec needs to be updated (I don't know ruby and don't really have an idea how to do it)

• Update code documentation if necessary. Not necessary (I think)

[iberianpig]

Hi @ces42, thank you so much for your nice pull request!

The failing RSpec, it should probably pass your tests with the following code

diff --git a/spec/lib/libinput_command_spec.rb b/spec/lib/libinput_command_spec.rb
index 949184f..3ecc8f1 100644
--- a/spec/lib/libinput_command_spec.rb
+++ b/spec/lib/libinput_command_spec.rb
@@ -132,8 +132,12 @@ module Fusuma
         end

         it 'should call dummy events' do
-          expect(Process).to receive(:spawn).with('dummy_debug_events',
-                                                  { out: @dummy_io, in: '/dev/null' }).once
+          th = double(Thread, pid: 'dummy pid')
+          expect(Open3).to receive(:pipeline_start).with(
+            ['dummy_debug_events'],
+            ['grep -v POINTER_ --line-buffered'],
+            { out: @dummy_io, in: '/dev/null' }
+          ).once.and_return([th])
           subject
         end
       end

[ces42]

I think this is ready now

[iberianpig]

@ces42 Thank you for your contribution!