Incorrect normalization when using named ranges in fitTo() and fitting several times

When using `fitTo` several times with the same named range, it seems that something goes wrong with the normalization of the PDF.
When plotting it, the normalization appears to be multiplied by N, where N is the number of fits performed with the same range.

See `test1.pdf`, where I've fitted once (looks fine) and `test2.pdf` after the second fit (normalization is broken).
These plots were generated with `bug_multiple_fits.py`.

The same problem seems to occur when I'm fitting a 2-dimensional model with a named range, only that it already happens after the first fit. See `bug_multiple_dimensions.py` for an example.