wwaEpv00 / wwaApco returning incorrect values?

[GregTheDev]

I've been experimenting with wwaEpv00 and wwaApco and seem to be getting different results depending on whether I use the current WWA NuGet package or source code.

Consider this input for wwaEpv00:
double jd1 = 2456122.0;
double jd2 = -0.374222409930591;

Using the NuGet package, SOFA C code (included in WWA source), liberfa I get these results:
pvh:
[0, 0]: 0.36358010848826561
[0, 1]: -0.87098205453928734
[0, 2]: -0.37758762761290249
[1, 0]: 0.015790881726547225
[1, 1]: 0.0055821572243759544
[1, 2]: 0.0024205553276501913

pvb
{double[2, 3]}
[0, 0]: 0.36128697138855309
[0, 1]: -0.87287395759745723
[0, 2]: -0.37843009179223269
[1, 0]: 0.01579627642463895
[1, 1]: 0.0055793863254509539
[1, 2]: 0.0024192303211176313

However if I run the same sample against the latest version of the WWA.Core source code I get different values:

pvh
{double[2, 3]}
[0, 0]: -1.4380401993025114
[0, 1]: 48.13081001851485
[0, 2]: 50.496624675136189
[1, 0]: -27.461656603117383
[1, 1]: -6.0043648152940641
[1, 2]: -3.2004034785635462

pvb
{double[2, 3]}
[0, 0]: -126.01563483201413
[0, 1]: -483.45362327412374
[0, 2]: -192.50804596930107
[1, 0]: -23.1220935093395
[1, 1]: 7.7476382322272448
[1, 2]: 2.7227051476989335

I see there were changes made to epv00 in Feb 2022, just not sure if it's expected that the result values have changed?

I noticed this because I was originally looking at wwaApco which seems to have a similar problem.

Given this input:
double jd1_tt = 2456122.0;
double jd2_tt = -0.3742224074074074;
double[,] ebpv = new double[2, 3] {
{ 0.36128697, -0.87287396, -0.37843009 },
{ 0.01579628, 0.00557939, 0.00241923 }
};
double[] ehp = new double[3] { 0.36358011, -0.87098205, -0.37758763 };
double x = 0.0012483996718683275;
double y = -2.4191277780194476e-05;
double s = 2.8198440882641448e-08;
double theta = 5.868498244632704;
double elong = -1.2915436464758039;
double phi = 0.7208209810736581;
double hm = 390.0000000015527;
double xp = 5.663587362550585e-07;
double yp = 1.973470449982449e-06;
double sp = -2.8551432482925797e-11;
double refa = 0.0;
double refb = -0.0;

When I invoke wwaApco using the NuGet package I get these results:
along: -1.2915436465043553
bm1: 0.99999999512048543
bpn: {double[3, 3]}
cphi: 0.75126413350351118
diurab: 0
eb: {double[3]}
eh: {double[3]}
em: 1.0165646807820918
eral: 4.5769545981283493
phi: 0
pmt: 12.530118487590943
refa: 0
refb: 0
sphi: 0.66000166796093673
v: {double[3]}
xpl: 2.0531311761394749E-06
ypl: -4.567827 644945285E-10

If I use the latest version of the source (as well as SOFA C code & liberfa) I get this:
along: -1.2915436465049144 // Last 4 digits different when compared to NuGet package
bm1: 0.99999999512048543
bpn: {double[3, 3]}
cphi: 0.75126413350351118
diurab: 0
eb: {double[3]}
eh: {double[3]}
em: 1.0165646807820918
eral: -1.706230709051797 // Different 4.5769545981283493 compared to -1.706230709051797
phi: 0 // -9.2559631349317831e+61 for SOFA, 2.999580946071453e-67 for liberfa
pmt: 12.530118487590943
refa: 0
refb: 0
sphi: 0.66000166796093673
v: {double[3]}
xpl: 2.0531311761393859E-06 // // Last 4 digits different when compared to NuGet package
ypl: -4.567827 0628706951E-10 // Different after the 6th digit when compared to NuGet package

So wwaApco seems closer to what I expect in the lastest source, but wwaEv00 seems worse.

Not sure if this is expected or if I'm doing something wrong?

Thanks
Greg