Add support for multiple postprocessing requests

[isaac091]

This reworks how translate configs are expected to look. Instead of specifying the postprocessing options inside of each translate request, there will be a separate 'postprocess' section at the top level which is a list of postprocessing requests that will be applied to all translate requests. A draft with no postprocessing applied is always saved. Once this is pushed, I will update the wiki with an example and explanation.

Example:

translate:
  - src_project: NASB
    books: MAT
  - src_project: NIV11R
    books: MAT
postprocess:
  - include_paragraph_markers: True  # only paragraph markers
  - include_paragraph_markers: True  # all markers
    include_style_markers: True
    include_embeds: True

It also changes the behavior of postprocess_draft.py slightly. Now, if none of the postprocessing options are toggled and the --experiment option is used, the script will apply any postprocessing requests in the experiment's translate config. Previously, if no postprocessing options were used, it would create a base draft, but that is no longer necessary because they are created by default, and I can foresee folks expecting to be able to use the translate config to configure this script. I will also update the wiki to reflect these changes.

This PR is ready to be reviewed, but not to push right away. I still need to get feedback on what the names of the output files should look like. Currently, the base draft will have the same file name as drafts do now, and any outputs with marker placement options will have a suffix, e.g. 41MAT_pse.SFM, that indicates if paragraph markers, style markers, or embeds were included.

Closes #746

---

This change is 

[benjaminking]

Reviewed 4 of 4 files at r1, 2 of 2 files at r2, all commit messages.
Reviewable status: all files reviewed, 3 unresolved discussions (waiting on @isaac091)

---

silnlp/common/postprocess_draft.py line 186 at r1 (raw file):

        if args.experiment:
            LOGGER.info("No postprocessing options used. Applying postprocessing requests from translate config.")
            with (config.exp_dir / "translate_config.yml").open("r", encoding="utf-8") as file:

I believe it's possible for a valid experiment folder not to contain a translate_config.yml file. If that's correct, you probably should have a check that the file exists.

---

silnlp/common/postprocess_draft.py line 189 at r1 (raw file):

                postprocess_configs = yaml.safe_load(file).get("postprocess", [])
            if len(postprocess_configs) == 0:
                LOGGER.info("No postprocessing requests found.")

Can you make this message a little more specific?

---

silnlp/common/postprocess_draft.py line 248 at r1 (raw file):

            else UpdateUsfmMarkerBehavior.STRIP
        )
        marker_placement_suffix = (

Reading through this, I'm thinking that this is becoming sufficiently complex to be worth factoring out a class with the postprocess_config logic. It could handle the mapping from the property names to the enum values as well, creating the file suffix, and creating the draft remarks so that you don't have to duplicate code between here and translator.py.