Are there certain proteins (CDSs) of bacteria which get expressed in transcriptome more than others?

Of course.

and goal is to identify pathogens. I'm interested if it is possible to identify those bacterias by their CDSs found in that transcriptome.

Go to rRNA, as far as I know is the way to go.

For example, is it possible to identify bacteria by the subset of its CDSs which are found in a sample?

Yes, but you need to know their genomes and have good transcriptome coverage to be capable of map SNPs.

Is there some database which holds which subset of CDSs must be found (active) in order to identify certain bacteria?

I don't know of any of this ones, but you can get active in-vivo genes from published transcriptomic data.