Re-integrate nukes into analytical outputs

[zaneselvans]

• Generation fuel table is split into nukes and non-nukes
• Some output tables use the generation_fuel_eia923 table and assume that it contains ALL the generation_fuel records. Which is now false. E.g. MCOE and net generation allocation. However, these dependent outputs don't realize that they are now working with incomplete data.
• One option: concatenate the gf and gfn tables and recreate the previously existing generation_fuel_eia923 output table.
• Another option, create 3 (no nukes, with nukes, combined).
• Concern: Are we going to just lose the generator level granularity that we've got in the nukes? They're not many of the records, but they are 20% of the overall generation. May want to address that. Should create an issue to fix this and integrate that granularity into the net generation allocation process etc.

Example from CG implemented in the DBCP context:

    # generate the generation fuel table & its nuclear counterpart
    # in order to SQUISH THEM together and reassigned this squished
    # version of the gen fuel tables into the cached pudl_out table
    # this way the subsequent outputs which use the gen fuel table
    # *cough cough* MCOE have the nuclear generators in there.
    gf = pudl_out.gf_eia923()
    # only add it in if the nuclear ID isn't in the columns
    # which indicates that these two tables have not been SQUISHED
    if 'nuclear_unit_id' not in gf:
        gfn = pudl_out.gfn_eia923()
        pudl_out._dfs['gf_eia923'] = pd.concat([gf, gfn])