Do not immediately abort the extraction if a seed chunk is invalid

[RyuzakiKK]

• Factor out writeChunk to a separate function

• Do not immediately abort the extraction if a seed chunk is invalid

  If the seed points to a RW location it may happen that some files change
  while we are in the middle of an extraction.

  When using the "InvalidSeedActionRegenerate" option, we can try harder
  and attempt to take the invalid chunks from the self seed or the store.
  If both of those fail too, then we abort the entire operation.

• Limit the maximum number of chunks is a single seed sequence

  If a seed is nearly completely equal to the output, we may end up with
  just a few SeedSequencer that are very long and others that are just
  one/a few chunks long.

  Because each SeedSequencer is handled by a goroutine (from a pool), we
  may reach a situation where the majority of the goroutines finish their
  operations and are just waiting for the longer SeedSequencer jobs to
  end.

  By limiting the maximum amount of chunks in the SeedSequencer, we will
  have jobs that are more balanced in term of amount of work.

• When taking a chunk from the self seed, immediately return

  If we were able to take a chunk from the self seed, there is no need to
  continue looking into the existing file and the store.

[RyuzakiKK]

I manually tested the situation where a seed changes while desync extract is in progress.

However I'd like to add some automated tests too. I'll probably look into it in the days.