fix(timeout): channels shouldn't block

as otherwise we would leak the fiber if there is a timeout as the channels would never receive

shard.yml

@@ -1,5 +1,5 @@
 name: tasker
-version: 2.0.6
+version: 2.0.7
 crystal: ">= 0.36.1"
 
 dependencies:

spec/tasker_spec.cr

@@ -188,25 +188,25 @@ describe Tasker do
   it "should pause and resume a repeating task" do
     sched = Tasker.instance
     run_count = 0
-    task = sched.every(40.milliseconds) { run_count += 1; run_count }
+    task = sched.every(80.milliseconds) { run_count += 1; run_count }
 
     begin
       tasks << task
 
-      sleep 50.milliseconds
+      sleep 100.milliseconds
       run_count.should eq(1)
 
-      sleep 40.milliseconds
+      sleep 80.milliseconds
       run_count.should eq(2)
 
       task.cancel
 
-      sleep 40.milliseconds
+      sleep 80.milliseconds
       run_count.should eq(2)
 
       task.resume
 
-      sleep 50.milliseconds
+      sleep 100.milliseconds
       run_count.should eq(3)
     rescue error
       puts "\nfailed cancel running tasks\n#{error.inspect_with_backtrace}"

src/tasker/timeout.cr

@@ -7,8 +7,8 @@ class Tasker
     end
 
     def execute!
-      success = Channel(Output).new
-      failure = Channel(Exception).new
+      success = Channel(Output).new(1)
+      failure = Channel(Exception).new(1)
 
       if @same_thread
         fiber = Fiber.new { perform_action(success, failure) }