Fail DB::Open() on WAL corruption

Summary:
This is a serious bug. If paranod_check == true and WAL is corrupted, we don't fail DB::Open(). I tried going into history and it seems we've been doing this for a long long time.

I found this when investigating t5852041.

Test Plan: Added unit test to verify correct behavior.

Reviewers: yhchiang, rven, sdong

Reviewed By: sdong

Subscribers: dhruba, leveldb

Differential Revision: https://reviews.facebook.net/D30597

db/corruption_test.cc

@@ -231,7 +231,9 @@ TEST(CorruptionTest, Recovery) {
   Check(100, 100);
   Corrupt(kLogFile, 19, 1);      // WriteBatch tag for first record
   Corrupt(kLogFile, log::kBlockSize + 1000, 1);  // Somewhere in second block
-  Reopen();
+  ASSERT_TRUE(!TryReopen().ok());
+  options_.paranoid_checks = false;
+  Reopen(&options_);
 
   // The 64 records in the first two log blocks are completely lost.
   Check(36, 36);

db/db_impl.cc

@@ -905,7 +905,7 @@ Status DBImpl::RecoverLogFiles(const std::vector<uint64_t>& log_numbers,
     std::string scratch;
     Slice record;
     WriteBatch batch;
-    while (reader.ReadRecord(&record, &scratch)) {
+    while (reader.ReadRecord(&record, &scratch) && status.ok()) {
       if (record.size() < 12) {
         reporter.Corruption(record.size(),
                             Status::Corruption("log record too small"));