I'd try Funannotate which is the most convenient annotation pipeline I know. While many public genome annotation are made with the Gnomon or Ensembl pipelines, it is hardly possible to install and run them locally without the support of the developer teams (possibly not even then). You would have to test if it likes your external Braker2 gffs but it shouldn't be a show-stopper.

See the documentation to funannotate annotate. Caveat: you still need to run InterProScan manually which is the heart of functional annotation.

If you need an alternative and want to submit your assembly to ENA, you can take a look at the snakemake annotation pipeline we are using in the EarthBio Genome project in Norway.

Running blast against NR seems like an overkill. You can use your CDS to blast against Uniprot (locally, i.e. download all plant-related entries and build a blast-db ) using blastx.

Alternatively, you can use HMMER3 on a translated protein fasta, which you can create e.g. via Emboss' transeq.

I prefer PANNZER2 (http://ekhidna2.biocenter.helsinki.fi/sanspanz/). Just upload a FASTA file with proteins, and PANNZER2 will give you a list of their descriptions.