Remittance theory and code need to be fixed

[rickecon]

The remittances theory and code need to be fixed. When we ran OG-PHL with calibrated remittances (see OG-PHL PR #39 and OG-PHL Issue #40), the steady state computation gave a resource constraint error. This suggests an error in the theory and how it is coded.

cc: @jdebacker

[jdebacker]

@rickecon I agree. Looking at the resource constraint equation, it appears that RM enters with the wrong sign.