I believe I've spotted a major flaw in your analysis concerning offsetting methane emissions. Your calculations arriving at a 50 million tonne per year increase in CO2e from methane leaks for the natural gas power generation used to back out coal-fired power starts with the entire 2012 natural gas consumption of the US electric power sector, of 9.1 trillion cubic feet per EIA stats. Yet that 9.1 TCF was used to generate over 1,200 million MWh (per EIA stats), not just the 160 million MWh you attribute to substitution of coal by gas. (EIA's analysis to which you link for validation attributes over 200 million MWh to fuel switching.) In fact, the relevant input to your methane offset calculation must be the increase in natural gas consumption for power generation from 2011 to 2012, or 1,563 BCF. By a simple ratio, that reduces your calculated methane offset from 49 million tonnes of CO2e to a little over 8 million tonnes.
You can test this by backing into a similar result using heat rates. 160 million MWh of power at an average heat rate of 9,000 BTU/kWh (averaging CCGT plants ranging from 6,000-8,000 BTU/kWh and peakers up to 12,000 BTU/kWh) yields around 1.4 TCF of incremental gas demand. Again, that's much less than the 9.1 TCF you used in your calculations.
Either way, counting only the assumed methane emissions attributable to the incremental natural gas consumption associated with the additional displacement of coal-fired electricity in 2012, rather than the assumed methane leakage for the entire US natural gas power generation fleet, results in a much smaller methane offset than you claim. That leaves a sizeable net emissions reduction benefit for gas, even if you assume that it also displaced a bit of hydro and nuclear power.