Fixing context of tasks when running pipeline

[Shahar-Y]

Fixes #1410

I noticed the agents ignore the context tasks given to them in the context field.
After thorough debugging I fund that the problem lays in the deep copy in crew.py (def copy(self) function) .
The problem was that when running the pipeline, it first copies the crews and then runs the copied crews.
When deep-copying the crew, the tasks are being copied, but they create another instance of the task in the context. That instance's output is not updated when the real task completes and therefore the tasks in the context always have an empty output field.

To solve it, I compared the keys of the original copied tasks with the tasks in the context of each task, and if they were the same, I replaced the context task with the original one. That way the context points to the task that is being run. It works perfectly now.

Proof it works:

Now the example in #1410 gives the correct answer:

[bhancockio]

Thank you so much @Shahar-Y for identifying the issue and submitting a solution. Your work was really helpful in pinpointing the problem and validating the fix!

I noticed a few additional linting issues and made some tweaks to the fix to align it with our coding standards. I’ll be closing out this PR in favor of this new PR #1416 , which incorporates your solution but with a slightly different approach.

Thanks again for your contribution, and I’ve credited you in the final PR! 🙌