Fix short SequentialFile reads that could drop WAL/MANIFEST data

[uadhran]

Summary

Fixes a data-loss risk when SequentialFile::Read() returns a successful short read before EOF (e.g. POSIX read() interrupted by a signal after partially filling the buffer). log::Reader previously treated any short read as EOF, which could silently drop MANIFEST or WAL data during DB::Open().

Changes

• util/env_posix.cc: PosixSequentialFile::Read loops until n bytes are read or EOF.
• util/env_windows.cc: Same retry loop for WindowsSequentialFile::Read.
• db/log_reader.cc: ReadPhysicalRecord accumulates partial block reads before treating EOF.
• include/leveldb/env.h: Document that implementations should only return short reads at EOF.
• db/log_test.cc: Add PartialReadSourceTest regression test; update StringSource mock to match real EOF semantics.

Testing

Built and ran on Linux (CMake Release):

./leveldb_tests --gtest_filter="*PartialRead*"          # 1/1 passed
./leveldb_tests --gtest_filter="LogTest.*:PartialReadSourceTest.*"  # 39/39 passed
./env_posix_test                                       # 7/7 passed
./leveldb_tests                                        # 210 passed, 2 skipped

Did not run Windows or macOS CI matrix locally.

Related: #1285

[google-cla]

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.