In addition to permeakra's comment, I want to add the following thoughts:
1) How were these bond lengths determined?
Were these values determined computationally? If so, what is the standard error in the computational method uses? In other words, is the difference between the two values within the error of the method, and thus the difference is not significant. Additionally, a computational method operates on single molecules "in the gas phase", and there might be just enough interaction between the hydrogen atom and one of the other oxygen atoms. This interaction would also explain the disparity in bond angles:
Were they measured experimentally, for example from x-ray crystallography? If so, then dissymmetry arising from a preferential crystal packing arrangement can lead to differences. In particular, in the solid state, the position of the hydrogen atom can make a big difference in preferential hydrogen bonding, either intermolecularly or intramolecularly. For units like the following hyrogen-bond dimers, the hydrogen bonding could lead to dissymmetry between the N-O bonds.
2) Bonds vibrate
They do, even in the solid state. Thus, an x-ray crystal structure has error its measurements as well. Again, is this difference in bond lengths within the experimental error?