regression: 1K RSA key upload length encoding

[martinpaljak]

What used to be 0x80
A>> T=1 (4+0145) 84D80001 91 73A180CE61E142BA3CD551AF4857309320A9E8026BBEFECE5B88C32C2B977C3A4CBC74E4A2E031C86591B96631C6C126930701F9C2DD0BBF6797BD43A9F450DE59BF2DC421842CAF06DAC74995BFAC572CC7AB873D79AB2778890EE32BE6BF85D16AA8B96E68E7537F97F7B787C2DFC8EB85D701F63FA8C7C14CA9D99275CC541385B9A003010001008928E4B3BA0B96FB 00

became 0x8180

A>> T=1 (4+0146) 84D80001 92 73A18180B1204569BA5FD20B7AC09782C6B88333257F5128D9FCCE8AF0E30434DFDA2538ED19E1A1EFE0FDC07DBEAB79AE650B684AF03CF53C318239DFB11EE42119B1B744DF513240C17FCCEB377432A521B7E35E2BD0BF43310EE68706B0D7480B9A3663CAAFC7DE873EFEC335C263913FD89F452A737E5D37632FD100227D7ACA1763A003010001004BD1A758E08E2660 00

And the relevant portion from GP spec:

The length field for ... is as defined for ASN.1 BER-TLV (see [ISO 8825-1]) except that the length 128 may also be coded on one byte as '80'.

Depending on the card, some like it one way some the other. See #374