ignorefpc correct usage and implementation

[smishr]

if ignorefpc
  @warn "assuming all weights are equal to 1.0"
  weights = ones(nrow(data))
end

If population size is not know or cant be estimated for some reason, ignorefpc parameter should provide usage without the (N-n)/N correction factor applied to mean/var etc. But I don't think that the way it is being done currently (above) is fully correct. Right now weights vector is made 1s, which is not quite ignoring fpc (although it does end up giving correct variances and means from svymean and svytotal)

This functionality and usage should be explored in more depth. If current implementation is found lacking, it could be removed and replaced. Although for what its worth, any survey that has sampled a substantial portion of the population would desire to use the fpc correction factor in their calculations.

[smishr]

For now, focus is on building a general survey design class that bypasses this issue entirely. So not going to focus resources at solving it right now