I have changed the simultaneous fit used for the single lepton bin from (pTrel,d0) to (pTrel,d0Significance).  The difference in yield is as follows:

TCE mu+:

N(h,d0Sig)=826.10983 +- 40.54608 ; N(h,d0)=954.24392 +- 48.73117

N(l,d0Sig)=896.89017 +- 40.54608; N(l,d0)=823.75608 +- 48.73117

TCE mu-:

N(h,d0Sig)=822.96071 +- 39.92599; N(h,d0)=901.60498 +- 47.26403

N(l,d0Sig)=842.03929 +- 39.92599; N(l,d0)=810.39502 +- 47.26403

CMUP mu+:

N(h,d0Sig)=438.87866 +- 30.63642; N(h,d0)=483.55751 +- 37.35530

N(DY,d0Sig)=215.57069 +- 65.82153; N(DY,d0)=183.67719 +- 67.78331

N(l,d0Sig)=297.55065 +- 66.95095; N(l,d0)=329.76530 +- 70.65134

CMUP mu-:

N(h,d0Sig)=421.02463 +- 29.85818; N(h,d0)=472.16470 +- 36.33199

N(DY,d0Sig)=181.33300 +- 56.34368; N(DY,d0)=151.93117 +- 56.44550
N(l,d0Sig)=328.64236 +- 58.58437; N(l,d0)=343.90413 +- 60.80432

CMX mu+:

N(h,d0Sig)=279.32750 +- 24.66394; N(h,d0)=306.06473 +- 30.01228

N(DY,d0Sig)=140.98705 +- 54.15454; N(DY,d0)=117.97464 +- 56.86518
N(l,d0Sig)=198.68545 +- 55.10031; N(l,d0)=213.96063 +- 59.11105

CMX mu-:

N(h,d0Sig)=272.94005 +- 24.27593; N(h,d0)=281.98879 +- 28.44901
N(DY,d0Sig)=231.44420 +- 55.78777; N(DY,d0)=229.34208 +- 58.18416
N(l,d0Sig)=93.61575 +- 53.50138; N(l,d0)=111.66913 +- 55.73253

Notice that the error on the fit using the impact parameter significance is always smaller, suggesting that the shapes are better separated in this variable.  All plots are below the break.