Incorrect size-1vertical dimension for some variables

TOTAL MOISTURE FLUX variables are created with a size-1 vertical dimension, when they shouldn't have one.

Example lookup header:

UM_m01s30i407_vn1308: 1/124 lookup headers
------------------------------------------
LBYR: 1950 LBMON: 1 LBDAT: 1 LBHR: 6 LBMIN: 0 LBDAY: 0 LBYRD: 1950 LBMOND: 1
LBDATD: 1 LBHRD: 0 LBMIND: 0 LBDAYD: 0 LBTIM: 11 LBFT: 6 LBLREC: 4915200
LBCODE: 1 LBHEM: 0 LBROW: 1920 LBNPT: 2560 LBEXT: 0 LBPACK: 0 LBREL: 3 LBFC: 0
LBCFC: 0 LBPROC: 0 LBVC: 0 LBRVC: 0 LBEXP: 0 LBEGIN: 94638592 LBNREC: 4915200
LBPROJ: 900 LBTYP: 0 LBLEV: 0 LBRSVD1: 0 LBRSVD2: 0 LBRSVD3: 0 LBRSVD4: 0
LBSRCE: 13081111 LBUSER1: 1 LBUSER2: 94638592 LBUSER3: 0 LBUSER4: 30407
LBUSER5: 0 LBUSER6: 0 LBUSER7: 1 BRSVD1: 0.0 BRSVD2: 0.0 BRSVD3: 0.0
BRSVD4: 0.0 BDATUM: 0.0 BACC: -99.0 BLEV: -1.0 BRLEV: 0.0 BHLEV: 0.0
BHRLEV: 0.0 BPLAT: 90.0 BPLON: 0.0 BGOR: 0.0 BZY: -90.046875 BDY: 0.09375
BZX: -0.0703125 BDX: 0.140625 BMDI: -1073741824.0 BMKS: 1.0

Leads to

>>> # CURRENTLY have spurious `dimension` dimensions:
>>> import umfive
>>> u = umfive.File('dz876a.pc1950jan')
>>> print(u)
dz876a.pc1950jan: <umfive.File: 2 data variables, 7 metadata variables>
    Data variables:
        UM_m01s30i407_vn1308: <umfive.DataVariable: UM_m01s30i407_vn1308, shape=(124, 1, 1920, 2560), dimensions=(time, dimension1, latitude, longitude)>
        UM_m01s30i408_vn1308: <umfive.DataVariable: UM_m01s30i408_vn1308, shape=(124, 1, 1920, 2560), dimensions=(time, dimension1, latitude, longitude)>
    Metadata variables:
        time: <umfive.DimensionScale: time, shape=(124,)>
        latitude: <umfive.DimensionScale: latitude, shape=(1920,)>
        bounds2: <umfive.DimensionScale: bounds2, size=2>
        latitude_bounds: <umfive.Variable: latitude_bounds, shape=(1920, 2), dimensions=(latitude, bounds2)>
        longitude: <umfive.DimensionScale: longitude, shape=(2560,)>
        longitude_bounds: <umfive.Variable: longitude_bounds, shape=(2560, 2), dimensions=(longitude, bounds2)>
        dimension1: <umfive.DimensionScale: dimension1, size=1>

when what we expect is no dimension1 dimension scale:

>>> # EXPECTED no vertical dimension:
>>> u = umfive.File('dz876a.pc1950jan')
>>> print(u)
dz876a.pc1950jan: <umfive.File: 2 data variables, 6 metadata variables>
    Data variables:
        UM_m01s30i407_vn1308: <umfive.DataVariable: UM_m01s30i407_vn1308, shape=(124, 1920, 2560), dimensions=(time, latitude, longitude)>
        UM_m01s30i408_vn1308: <umfive.DataVariable: UM_m01s30i408_vn1308, shape=(124, 1920, 2560), dimensions=(time, latitude, longitude)>
    Metadata variables:
        time: <umfive.DimensionScale: time, shape=(124,)>
        latitude: <umfive.DimensionScale: latitude, shape=(1920,)>
        bounds2: <umfive.DimensionScale: bounds2, size=2>
        latitude_bounds: <umfive.Variable: latitude_bounds, shape=(1920, 2), dimensions=(latitude, bounds2)>
        longitude: <umfive.DimensionScale: longitude, shape=(2560,)>
        longitude_bounds: <umfive.Variable: longitude_bounds, shape=(2560, 2), dimensions=(longitude, bounds2)>

Start at the umfive.File loading path and reproduce the issue with dz876a.pc1950jan, checking how TOTAL MOISTURE FLUX lookup headers become dimensions. Confirm that the two example variables no longer get a size-1 vertical dimension or dimension1 scale, while time, latitude, longitude, and their bounds remain unchanged.