Anyway what I found the first time around is you get twice as much energy out of NG by burning it, rather than combining it with a reactant like oxygen. ...
You got that backwards, Sunking... Fuel cells are at least twice as efficient at converting chemical energy into electrical than an internal combustion engine. Also, both systems combine a fuel with oxygen to release the energy of formation (enthalpy) - the fuel cell uses a catalyst to liberate the energy as electrons while a heat engine liberates the energy and is subject to Carnot's Rule.
A third means of combining the two reactants - e.g., fuel and oxygen - which is even more efficient is nuclear fusion... needless to say, that's probably not coming to a store near you anytime soon.
I am a bit suspicious of the 250,000 miles - last time I had concrete evidence the membranes lasted less than a year - even less if the fuel was not really pure
Depends on the fuel cell. The higher temperature types like Solid Oxide and Molten Carbonate have very long operational lives, don't need exotic noble metal catalysts and are very tolerant of impurities in the fuel/air stream feeding them, but they are - as you might guess - only suitable to fixed installations.
Lower temperature cells like Alkaline (used in space flight) and Polymer Electrolyte Membrane (the type mentioned here) are much less tolerant of impurities in the feedstock, particularly sulfur compounds. The PEM type also needs to be kept at the proper humidity level to maintain ionic conductivity of the special sulfonic acid-polymer membrane (e.g. - DuPont's Nafion) without flooding the gas passages with liquid water. Once you solve all these physical problems the PEMFC is remarkably robust and should pump out power for decades - 250k miles should be trivial in such a case.
PEMFCs would make an almost ideal range extension generator in a series hybrid if it weren't for one thing.... their cost. PEMFCs require very active catalysts (e.g. - precious metals) because of their low operating temperature, the plastics used for the ion exchange membrane are difficult to manufacture and the electrode materials have to withstand high humidity in the presence of a potential difference (ie - subject to galvanic corrosion) so they are inevitably made of expensive materials like machined graphite.
All of this issues are potentially solvable, we just ain't there yet.