Revert "Revert of [turbofan] greedy: heuristic for memory operands

Revert "Revert of [turbofan] greedy: heuristic for memory operands
(patchset #2 id:40001 of https://codereview.chromium.org/1306823005/ )"

This reverts commit f68cd33bdd342819abd5b091591ac78383c2e79d.

BUG=

Committed: https://crrev.com/6009697eb22327a2b74f7153251c09ac6d042896
Cr-Commit-Position: refs/heads/master@{#30509}

[Mircea Trofin, 2015-09-01 10:46:52]

mtrofin@chromium.org changed reviewers:
+ bmeurer@chromium.org, jarin@chromium.org

[Jarin, 2015-09-01 11:57:32]

lgtm, although I do not fully understand why there is the call to
FindOptimalSplitPos after we find a valid split position.

[Mircea Trofin, 2015-09-01 12:31:14]

The CQ bit was checked by mtrofin@chromium.org

[commit-bot: I haz the power, 2015-09-01 12:31:20]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1314423006/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1314423006/20001

[Mircea Trofin, 2015-09-01 12:33:00]

On 2015/09/01 11:57:32, Jarin wrote:
> lgtm, although I do not fully understand why there is the call to
> FindOptimalSplitPos after we find a valid split position.

There are many positions we may split. Before, we were picking the one just
before the first use needing
a register. That may not be optimal, because it may happen the first such use is
in a loop. The optimal
position logic attempts to find an alternative, valid split position, outside of
a loop.

[commit-bot: I haz the power, 2015-09-01 12:33:06]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2015-09-01 12:33:17]

Patchset 2 (id:??) landed as
https://crrev.com/6009697eb22327a2b74f7153251c09ac6d042896
Cr-Commit-Position: refs/heads/master@{#30509}