1. If suprnova were selfish-mining, it would benefit all contributors to the pool. A pool can steal from its contributors in many ways, but straightforward selfish-mining is not one of them.
2a. The bad luck exhibited by btg.suprnova.cc is extraordinary. According to the statistics shown at
https://btg.suprnova.cc/index.php?page=statistics&action=blocks, lifetime, enough shares have been submitted to find almost 2500 blocks (2480 as I write), with a standard error of 50 blocks. In fact, only 1505 blocks have been found, a variation of 20 standard errors. Luck that bad by chance is more unlikely than finding two blocks on your first two shares submitted.
2b. There were problems with some of the pool software, including code that extracted a secret fee for the benefit of one of the developers. But even if we look at just the last week of mining, we see enough shares to mine 959 blocks, with a standard error of 31 blocks... and only 633 have actually been mined. That's over 10 standard errors of bad luck. So even if pool software problems caused the earlier apparent bad luck, those should have been fixed for more than a week, yet the bad luck continues.
3. Having over 50% of the blocks of a chain mined by a single entity is brittle. All of us who own, mine, receive or spend BTG would be better off if suprnova's share of the mining were substantially lower than it is.