Improve PMM install behavior in cci flow

[samanthashain]

When using the flow cci flow run pmm:install_prod, the default behavior is to create the pmm metadata as well as about 1,000 sample records that are stored in a .sql file in the pmm repo.

Since we are creating scratch orgs FOR new data (usually), we should probably figure out how to wipe those records or install pmm without records.

We can use cci task run delete etc but we might want to run recipes that have a dependency on Programs and Services existing.

A group should determine the appropriate behavior for installing pmm and then update the cumulusci.yml file accordingly.

[emilyhirsch-CoastalCloud]

Additionally, we should have a CCI flow that is similar to the NPSP_scratch flow so that we can install these dependencies in a more consistent way.

[jenncarneiro]

Updating the labels as this seems to be an improvement and not a bug, and it is for a PMM install flow not a recipe.