The classic way of dealing with these unpredictable effects in the analog world is :
1.) Build your circuit with the calculated components on a breadboard and do a thorough analysis of its behavior vs supply voltage, temperature, humidity, etc. variations until you find differences between the expected end the real circuit behavior. Identify the components or sub-circuits which are responsible for these deviations and either replace these by better suited ones before restarting the analysis or think about making one or more components variable to obtain the expected behavior. Consider that manual calibration is a huge cost factor in production and should be avoided if possible, through intelligent almost self-adjusting or self-compensating circuit design. Having done EE studies might be helpful here.
2.) Design a PCB with respect to thermal management and order a small quantity for prototyping. Build a few (10 or so) prototypes and analyze/observe these thoroughly as before, simulating all use cases and environmental conditions. Check if you can always obtain the expected results, if needed through one-time calibration of the variable components. If you need to recalibrate several times a day/week, it's poor design and you should go back to step 1.
3.) From these processes, you end up with an optimized circuit and PCB design and you might finally consider starting mass production.
There is no sense in choosing components with similar tolerances without deeper thinking and analysis. Lower tolerance means often higher cost, so the rule for tolerances is as low as possible, as high as needed. When it comes for example to a pull-up resistor for a switch, tolerances will not matter, take the cheapest one you can get. But when it comes to the generation of reference voltages, i.e. for precision analog signal processing in the uV domain in EEG or EMG applications, the more expensive 0.1% resistors might make things easier. Taking all that into account requires lots of trial and error at the beginning, but you'll acquire the needed experience over the years until you come to a point where you might do a big part of that work in mind, or where you look at some schematics and automatically identify critical components.