Why bacterial system is still used for protein production for some protein knowing that it doesn't have a system to support post translational modification?
I suppose it's easier to work with an established protein expression system than to engineer a new system just to ensure the background will provide an ideal mix of enzymes for the desired end product. That being said, if the desired post-translational modification and the responsible enzyme are already known, it often makes sense to just engineer the existing expression system to contain the enzyme (assuming it's active, of course).
The medium costs and the infrastructure are way cheaper.
1L of medium to grow bacteria is way cheaper than 1L of a medium for human or insect cells. Bacteria are also easier to lyse than yeast for example. An exception would be yeasts that secrete the protein of interest, but this does generally only work well for proteins that are anyway secreted...
So most people try first with bacteria.
If it is known that PTMs are important for the question that is to be answered with the recombinant protein or the bacterial expressed protein is not active usually then people switch to the other systems.
Also, for bacteria you just need medium, a flask and a shaker to get a descent amount of cells. For insect or human cells, you need an incubator with CO2 for certain medium formulations (pH control of the medium), and/or special flasks to ensure that enough O2 is present when the cells should grow dense, making it rather expensive for the same amount of recombinant protein.
In addition to the answers already provided, there are cases where the post-translational modification is not desired. For instance, if you have a glycosylated protein and you want to make antigen for the production of specific antibodies, you may wish to have a non-glycosylated protein, because glycans are present on many proteins and often share similar structures. Antibodies against the glycan moiety may thus recognise several other proteins as well, instead of just your protein of interest.