For all the common uses of a spectrophotometer in biology, any commercial one should be sufficient. For all the routine stuff, the quality of the spectrophotometer is usually of negligible importance.
There are no general rules on what errors and resolution are acceptable, that always depends on what you're measuring. The error should be significantly smaller than size of the effect you're looking at, otherwise you get no useable data.
There are many sources of error with biological systems, the spectrophotometer is really one of the last ones you would think of. Simply the error of pipetting is likely far larger than the error of the spectrophotometer, and differences in the preparation of your biological samples might introduce an even larger error. Unless you're doing something beyond the routine uses of UV/Vis spectrometry, you probably don't need to worry about this.