Allow Prigent streams to be off in Leung_2023 dust emissions for Paleo work

When #1897 comes in the Prigent Roughness streams will be required for the Leung_2023 option. In order for this to make sense in a Paleo context we will need the ability to turn the streams off. This would be setup this way in the 1850_noanthro_control use-case, so used in the I1850Clm50BgcNoAnthro compset (and other NoAnthro compsets).

A suggestion on how to do this from @dmleung the author is:

"Add option to turn off Prigent data in Leung? Or have a constant value dataset? Danny believes that we set the baresoil landcover fraction to zero and the vegetation landcover fraction to be one."

Another way would be to assign a global average roughness for baresoil that's used in place of the dataset. This would be for example important in order to assign something to any gridcells that are 100% baresoil.

We'll do this after Leung_2023 first comes in, to facilitate Leung_2023 coming in as fast as possible. And we should consult with the Paleo group at that time to make sure it makes sense for them.

Definition of done:

- [ ] Allow it be turned off and a constant value used for testing (suggest 0.1)
- [ ] Remove the stub code under src/unit_test_stubs/share_esmf and use the normal one (reduces maintenance of test code by lowering duplication)
- [ ] Make sure the option is tested especially in the PF unit testing
- [ ] Work with the Paleo group to plan on how should be handled (additional dataset), ability to set the constant value used?
- [ ] Implement what the Paleo group needs