For one thing, you presumably mean "flesh-eating bacteria". Viruses don't "eat" in anything but a trivial sense.

Could one direct a bacterium at a specific cancer tissue? In theory one could, but it would be a huge ask. The bacterium would have to be deadly to the cancer without being deadly to the rest of your body. Certainly some bacteria will attack particular tissues, but mainly because of how easily they can invade them with whatever enzymes they have and in the particular conditions they live in the body.

Could there be bacteria already that attack tumours? Very likely, but probably only when the tumour is large and well-defined. Against such tumours oncologists don't need much help; it is the subtle, moving, spreading cells that mock our treatments.

Cancers can occur pretty well anywhere in the body, and they resemble your own tissues very closely (mostly they are your tissues anyway!) so it is hard to imagine how to direct a germ against them and nothing else, and keep it happening till all the cancer cells are dead.

I cannot say that we never will know our tissues in enough detail to know how to tell the tumours from the good tissues, nor that we never will know bacteria well enough to set them on just such a tumour, but I also cannot help thinking that by the time we know such things, it would be easier to set our own cells on the tumour; they have quite advanced recognition and killing capabilities already.

To create a virus that only attacks cancer, you need a surface marker that only occurs on cancer cells and which is recognized by the virus. This might be technically possible, but has some drawbacks. It would only work if the cancer had such a marker, you might have to develop a different viral strain for each type of cancer and there is always the worry that a mutation would lead the virus to attack all the cells in your body. This combination of price and danger probably makes looking for such a cure unattractive to the health industry.

A related approach is being used: creating antibodies that only recognize and bind to cancer cells, using a similar idea, that they will recognize and bind to specific proteins only found on the cancer. The initial results are promising but, since it is specific for each person, the cost is currently prohibitive as a general treatment