Fix edge_steps unique values

[avaly]

For ColumnCount = 8, edge_steps contain a double value for 0.5, which makes windows cycle only between 0.125 and 0.5 of the screen width.

Before:

DEBUG: edge_steps (0.5, 0.125, 0.25, 0.375, 0.5, 0.625, 0.75, 0.875)

After:

DEBUG: edge_steps (0.125, 0.25, 0.375, 0.5, 0.625, 0.75, 0.875)

[coveralls]

Coverage remained the same at 32.633% when pulling e5f65b8 on avaly:fix/edge_steps into 31b8144 on ssokolow:master.

[ssokolow]

While I appreciate you offering up a PR this quickly, I don't want that sorted in there.

While I could quickly whip up a solution which throws out duplicates while preserving ordering, the actual proper solution is just to track where in that sequence it is without getting stuck... and, if all goes well, that change will be complete by the end of today in order to fix #20 and to prepare to fix #24, #10, and #30.

I'll leave this open as a reminder to test this situation once it's done.

[ssokolow]

I wound up spending more time refactoring the code than I expected, so it'll probably be tomorrow instead.

[ssokolow]

There. Fixed. This also fixes #20 in that window size hints can't interfere either.

You might get weird layout bugs if if you try to tile something which imposes geometry restrictions like a video player, but at least the cycle won't get stuck. (I'd have actually had this ready hours ago but I got side-tracked working on the code to fix that other problem.)

If you want to track the progress on that follow issue #87.