there does not exist something as 'the recommended parameters for RNA-Seq' , so do not worry about it.

the only important parameter is what your experiment/analysis is like (or what you hope to get out this experiment)

Just leave them as they are and run your analysis with the complete set/read length. I fear your will hinder the downstream analysis more than that you will improve it by starting to fiddle with your input data

Don't do that, the longer the reads the better for RNA-seq as greater length makes alignment more reliable. Continue with the data as-is.

If you do need to trim the data down (which may be needed if you have pre-existing data that was sequenced at 75 bp and you want to do a comparative analysis) then you can use reformat.sh from BBMap suite.

reformat.sh -Xmx4g in=input.fastq.gz out=trimmed.fastq.gz forcetrimright=75 (# for single end reads)

reformat.sh -Xmx4g in1=R1.fastq.gz in2=R2.fastq.gz out1=R1_trim.fastq.gz out2=R2_trim.fastq.gz forcetrimright=75 (# for paired-end reads)

If this does not apply then you already have recommendations from others to continue as is.