Standard theory of evolution: organisms that spend energy on features that do not contribute to their fitness will be at a disadvantage in the long run compared to others of otherwise similar type that do not. The others will have more energy to devote to reproduction.

Citing notable research might help me better understand your claim; it is very possible that I'm missing something, but that's not how evolution works to my knowledge, adoption of substantially new genetic features is rapid, but refinements or removal of unrelated genetic code is slow if it is not impacted by the substantive mutations.

Source: http://www.columbia.edu/~gap2118/papers/nature13827.pdf

Resistant bacteria usually become resistant by giving up a metabolic pathway.

I wish I could have read more than the abstract, but maybe this [1] ... I would especially like to know what do they mean by "Unfortunately, the available data suggest that the rate of reversibility will be slow at the community level"

[1] http://www.nature.com/nrmicro/journal/v8/n4/pdf/nrmicro2319....

From that same paper:

> In hospitals, both modelling and analysis of the correlations between antibiotic resistance and variation in antibiotic use show that alterations in antibiotic use can cause rapid changes (in the order of days to months) in the frequency of resistance. By contrast, when the fitness cost of resistance is the main driving force behind its reversal, the rate of change is expected to be much slower (months to years).

I think what that means is that hospitals are able to create more selective pressure by rotating through different kinds of antibiotics.