Consider clearing errors at the beginning of a try block

Right now nested try blocks within catch will never trigger:

pipeline:
  processors:
    - try:
      - resource: foo
    - catch:
      - try:
        - resource: bar
        - resource: baz

So if foo fails neither bar nor baz would execute. This is because the error context is carried into the processors of the catch block and aren't cleared until the end so that the error message can be referrenced. A work around is to add a nested catch before the try block:

pipeline:
  processors:
    - try:
      - resource: foo
    - catch:
      - catch: []
      - try:
        - resource: bar
        - resource: baz

This clears the error before the try block but looks odd and ugly.

Reproduce the nested YAML pipeline with try, catch, and the foo, bar, and baz resources. Trace how the error context moves through these pipeline processors and determine how clearing it at the start of a try should affect nested execution. Done means bar and baz run after foo fails without requiring an empty nested catch, while the original error remains available for the catch block.