Fix Progress Manager Bars and Remove Duplicate execute_plan() Logic

[mdr223]

The progress managers have some small discrepancies with reality and duplicated code in a couple places, from earlier discussion:

• In this commit and this commit, which initially added progress bars to no sentinel execution:
  • The execute_plan() method was copied from each PlanExecutor, placed in the respective Execution class, and modified to include progress bars
    • (I also see -- particularly for parallel pipelined execution -- that code improvements were made)
• I can think of a lot of valid reasons the copying may have been done; my only concern is that now:
  • We have two execute_plan() methods floating around for each plan executor
  • We have a lot of duplicated code
• If possible, I'd love to come to a consensus on how best to eliminate the duplication
  • (this may include refactoring the abstraction between ExecutionEngine and PlanExecutor, because it is a bit clunky at the moment)

Small notes: I do not want to let the perfect be the enemy of the good, and I am very happy that we now have some progress monitoring!

• That being said, in the presence of aggregates, limits, and one-to-many converts, total_work_units = total_items * total_ops will not always be true
  • (I'm not sure how important this is from a correctness perspective, but thought I'd point it out)
• Our outer op progress tracker also does not really translate well for pipelined execution (since it proceeds one record at a time as opposed to one op at a time), so we may want to flip the inner vs. outer progress bars there
• So overall -- very glad this is here! -- but we may want to tweak some more