[sql] Disable memory-mapping under sql::Recovery.

[sql] Disable memory-mapping under sql::Recovery.

Recovery may be used to recover files with I/O errors, which would not
interact well with memory-mapped I/O.

[AFAICT, SQLite doesn't use memory-mapped I/O on temporary files, but as
of yet I don't see why this might be.]

BUG=551110
Committed: https://crrev.com/2f3a814be066d8c02ba8b82f003cd98799165882
Cr-Commit-Position: refs/heads/master@{#358083}

[Scott Hess - ex-Googler, 2015-11-04 23:34:58]

shess@chromium.org changed reviewers:
+ rmcilroy@chromium.org

[Scott Hess - ex-Googler, 2015-11-04 23:34:59]

I was working on I/O error handling WRT sql::Recovery, and it occurred to me
that mmap could be a problem, there, so making the contract explicit seems in
order.

[Scott Hess - ex-Googler, 2015-11-04 23:36:42]

https://codereview.chromium.org/1414563010/diff/1/sql/connection.cc
File sql/connection.cc (right):

https://codereview.chromium.org/1414563010/diff/1/sql/connection.cc#newcode1677
sql/connection.cc:1677: 
Hmm, and I moved the stanza down here because it seemed like the thing most
sensitive to database problems should come as late as possible to give other
statements a chance to flush out problems.

[rmcilroy, 2015-11-05 11:21:58]

lgtm

[Scott Hess - ex-Googler, 2015-11-05 18:06:07]

The CQ bit was checked by shess@chromium.org

[commit-bot: I haz the power, 2015-11-05 18:06:18]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1414563010/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1414563010/1

[commit-bot: I haz the power, 2015-11-05 18:11:16]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2015-11-05 18:11:56]

Patchset 1 (id:??) landed as
https://crrev.com/2f3a814be066d8c02ba8b82f003cd98799165882
Cr-Commit-Position: refs/heads/master@{#358083}