NRCHOP=3 not supported, but no trap or warning

[aprovost-usgs]

In a variation on ex01a that Joe generated using FloPy, he reported a problem with mass balance errors and backward-tracked particles not terminating properly at a recharge boundary. We determined that the problem stems from the fact that MODPATH doesn't support NRCHOP=3 (recharge to the highest active layer); it apparently simply skips over that budget information and doesn't die or issue any warning. Joe tracked it to these lines in pr_LoadTimeStep:

         else if(header%Method .eq. 3) then
             ! Not yet supported
         else if(header%Method .eq. 4) then

[aprovost-usgs]

MODPATH was unable to handle recharge to the highest active layer - the corresponding flows were not read from the budget when stored using budget method 3 (as when NRCHOP=3 for RCH in MF2005), and particles could fail to terminate properly in backward-tracking runs. The code is now updated to handle recharge to the highest active layer. In the process, also fixed an apparent bug (related to double vs single precision) that could result in failure to identify a cell as dry. The updated source code has been posted (commit a502e7f), and we're working on a USGS release that includes this update.