Prevent ghost breakpoints by making sure we can't have dangling tracepoints

Any time we disable a breakpoint or re-set the `tp` instance variable,
we need to make sure any previous trace points are disabled or else they
might continue to suspend execution even after breakpoint removal

Author: vinistock

lib/debug/breakpoint.rb

@@ -162,6 +162,11 @@ def initialize path, line, cond: nil, oneshot: false, hook_call: true, command:
     def setup
       return unless @type
 
+      # There are scenarios where `setup` is invoked more than once. If we don't disable any possible previous trace
+      # point events, then we will update the `@tp` reference with a new instance but the previous one will still be
+      # activated, leading to the debugger suspending in breakpoints that were previously removed
+      disable
+
       @tp = TracePoint.new(@type) do |tp|
         if @cond
           next unless safe_eval tp.binding, @cond

lib/debug/session.rb

@@ -1747,11 +1747,15 @@ def on_load iseq, src
         end.each do |_key, bp|
           if !bp.iseq
             bp.try_activate iseq
-          elsif reloaded
+          elsif reloaded && (iseq.first_lineno..iseq.last_line).cover?(bp.line)
             @bps.delete bp.key # to allow duplicate
-            if nbp = LineBreakpoint.copy(bp, iseq)
-              add_bp nbp
-            end
+
+            # When we delete a breakpoint from the @bps hash, we also need to deactivate it or else its tracepoint event
+            # will continue to be enabled and we'll suspend on ghost breakpoints
+            bp.delete
+
+            nbp = LineBreakpoint.copy(bp, iseq)
+            add_bp nbp
           end
         end
       else # !file_path => file_path is not existing