Dear OP, I'm going through the exact same process right now!
I was also a little skeptical about the number of transcripts in the Rat transcriptome, as opposed to the much larger number in the mouse one. However, I believe colindaven's answer explains it: not so many groups use rat as a model, so less information is known on its transcriptome.
As for genomax's answer, that was also where I obtained my transcriptome to index Salmon (does this happen to be the same reason you have to need this transcriptome?). In the stats link genomax posted, it says the rat transcriptome has 41,078 transcripts. However, the "Rattus_norvegicus.Rnor_6.0.cdna.all.fa" file has only 31,715 seqs. If you look into the non-coding RNA file (here: ftp://ftp.ensembl.org/pub/release-92/fasta/rattus_norvegicus/ncrna/Rattus_norvegicus.Rnor_6.0.ncrna.fa.gz), it has 9,331 seqs. Together, cdna and ncrna files amount to 41,046, which is roughly the number of transcripts said to be present in the rat transcriptome. I believe this could be how ensemble got to the number on the stats page of the rat genome/transcriptome.
You will also find an abinitio transcript file in the ensemble ftp (ftp://ftp.ensembl.org/pub/release-92/fasta/rattus_norvegicus/cdna/Rattus_norvegicus.Rnor_6.0.cdna.abinitio.fa.gz) which has 59,821 seqs. I do not know how this file relates to the others, so if anyone could help clear that up it would be great!
All in all, I used the shorter cdna fasta (31,715 transcripts) to index Salmon, but have also built indexes using the abinitio fasta and the concatenated file between cdna and ncrna. I'll do further anlyses running salmon with each one to see where I get to.
Best,
Thomaz