Make "tag_on_failure" a common filter setting

The `tag_on_failure` setting for the grok filter is very helpful with longer pipelines containing multiple grok filters, making it easier to locate the error in the pipeline, and more robust error handling.

For example, if a pipeline contained multiple [json filters](https://www.elastic.co/guide/en/logstash/current/plugins-filters-json.html), the ambiguous `_jsonparsefailure` is informative, but not entirely helpful.

No files or tests are named. Start by locating the grok and json filter implementations and comparing how filter settings and failure tags are handled; clarify the intended common setting and identify the relevant test entry points before defining completion.

Written by the indexing model from the issue text.