message when deleted inside retry processor, does not apply

Seeing something abnormal with retry processor, if the message was deleted in one of the retry processors and using a sql_insert output, was not expecting execution of sql_insert. But sql_insert fails with a error. Here is a sample pipeline that fails.

input:
  generate:
    interval: 1m
    mapping: |-
      root = { "p1": "test val" }
pipeline:
  processors:
  - retry:
      backoff:
        initial_interval: 500ms
        max_interval: 10s
        max_elapsed_time: 0s
      processors:
        - log:
            message: 'incoming msg ${! content() }'
        - mapping: root = deleted()
output:
  sql_insert:
    driver: postgres
    dsn: postgres://${DB_USER}:${DB_PASSWORD}@${DB_HOST}:5432/${DB_NAME}?sslmode=disable
    conn_max_idle_time: 1m
    table: test_table
    columns:
      - p1
    args_mapping: |
      root = [
        this.p1
      ]
    init_statement:
      CREATE TABLE IF NOT EXISTS test_table (
        p1 varchar
      ) 

and console output:

INFO incoming msg {"p1":"test_value"}              @service=benthos custom_source=true label="" path=root.pipeline.processors.0.retry.processors.0
INFO Output type sql_insert is now active          @service=benthos label="" path=root.output
ERRO Failed to send message to sql_insert: insert statements must have at least one set of values or select clause  @service=benthos label="" path=root.output
INFO incoming msg {"p1":"test_value"}              @service=benthos custom_source=true label="" path=root.pipeline.processors.0.retry.processors.0
ERRO Failed to send message to sql_insert: insert statements must have at least one set of values or select clause  @service=benthos label="" path=root.output
INFO incoming msg {"p1":"test_value"}              @service=benthos custom_source=true label="" path=root.pipeline.processors.0.retry.processors.0

Could workaround it by deleting message in a pre-processor to sql_insert, like below:

output:
  sql_insert:
    driver: postgres
    dsn: postgres://${DB_USER}:${DB_PASSWORD}@${DB_HOST}:5432/${DB_NAME}?sslmode=disable
    conn_max_idle_time: 1m
    table: test_table
    columns:
      - p1
    args_mapping: |
      root = [
        this.p1
      ]
    init_statement:
      CREATE TABLE IF NOT EXISTS test_table (
        p1 varchar
      )
  processors:
    - mapping: |-
        root = if this == null {
          deleted()
        }

Reproduce the supplied pipeline with the retry processor and sql_insert output, then trace how a deleted message moves from the retry processors to the output. Confirm the behavior against the PostgreSQL example; done means a message deleted inside retry processing does not cause sql_insert to execute.