kedro-telemetry: Improve performance by switching to after_command_run

[ankatiyar]

Description

Follow up on kedro-org/kedro#1476

From profiling various Kedro CLI commands, it turns out that a good chunk of the time is spent sending information to Heap in the before_command_run hook. This issue is to improve things on kedro-telemetry side.

Context

kedro-org/kedro#1476 (comment)
kedro-org/kedro#1476 (comment)

Possible Implementations

• Move to after_command_run hook, but this does not get executed if an exception occurs during the kedro run etc. The hook could be changed on the Kedro side to get executed even if the command fails.
• Try asynchronous sending of Heap data but this might be tricky since it's still an open request on pluggy support async hook calls pytest-dev/pluggy#320

Also to consider

There's some project hooks executed by kedro-telemetry too - we could figure out how to improve performance by not having to send this data twice

[ankatiyar]

• kedro-plugin side solution already in feat(telemetry): Combine Telemetry hook to send heap event once #766
• Kedro side solution kedro-telemetry: Improve performance by switching to after_command_run kedro#4014