Known failing test for Tile.getSites() result different to Vivado

[eddieh-xlnx]

On VU3P:

get_sites -of [get_tiles LAG_LAG_X30Y250]
WARNING: [Vivado 12-3731] No sites matched '*'

On VU5P:

get_sites -of [get_tiles LAG_LAG_X30Y250]
LAGUNA_X6Y140 LAGUNA_X6Y141 LAGUNA_X7Y140 LAGUNA_X7Y141

whereas RapidWright currently returns sites for all four testcases (the first three of which is incorrect).

Discovered during the analysis of #738 which uses the VU3P (not that nextpnr should place a regular FDRE onto a Laguna BEL, even if there was a valid site)