chore(sqlite): use normalised tables for file names and versions (#10383)

This changes the files table to use normalisation for the names and
versions. The idea is that these are often common between all remote
devices, and repeating an integer is more efficient than repeating a
long string. A new benchmark bears this out; for a database with 100k
files shared between 31 devices, with some worst case assumption on
version vector size, the database is reduced in size by 50% and the test
finishes quicker:

    Current:
        db_bench_test.go:322: Total size: 6263.70 MiB
    --- PASS: TestBenchmarkSizeManyFilesRemotes (1084.89s)

    New:
        db_bench_test.go:326: Total size: 3049.95 MiB
    --- PASS: TestBenchmarkSizeManyFilesRemotes (776.97s)

The other benchmarks end up about the same within the margin of
variability, with one possible exception being that RemoteNeed seems to
be a little slower on average:

                                          old files/s   new files/s
    Update/n=RemoteNeed/size=1000-8            5.051k        4.654k
    Update/n=RemoteNeed/size=2000-8            5.201k        4.384k
    Update/n=RemoteNeed/size=4000-8            4.943k        4.242k
    Update/n=RemoteNeed/size=8000-8            5.099k        3.527k
    Update/n=RemoteNeed/size=16000-8           3.686k        3.847k
    Update/n=RemoteNeed/size=30000-8           4.456k        3.482k

I'm not sure why, possibly that query can be optimised anyhow.

Signed-off-by: Jakob Borg <jakob@kastelo.net>

internal/db/sqlite/sql/schema/folder/20-files.sql

@@ -25,15 +25,27 @@ CREATE TABLE IF NOT EXISTS files (
     device_idx INTEGER NOT NULL, -- actual device ID or LocalDeviceID
     sequence INTEGER NOT NULL PRIMARY KEY AUTOINCREMENT, -- our local database sequence, for each and every entry
     remote_sequence INTEGER, -- remote device's sequence number, null for local or synthetic entries
-    name TEXT NOT NULL COLLATE BINARY,
+    name_idx INTEGER NOT NULL,
     type INTEGER NOT NULL, -- protocol.FileInfoType
     modified INTEGER NOT NULL, -- Unix nanos
     size INTEGER NOT NULL,
-    version TEXT NOT NULL COLLATE BINARY,
+    version_idx INTEGER NOT NULL,
     deleted INTEGER NOT NULL, -- boolean
     local_flags INTEGER NOT NULL,
     blocklist_hash BLOB, -- null when there are no blocks
-    FOREIGN KEY(device_idx) REFERENCES devices(idx) ON DELETE CASCADE
+    FOREIGN KEY(device_idx) REFERENCES devices(idx) ON DELETE CASCADE,
+    FOREIGN KEY(name_idx) REFERENCES file_names(idx),
+    FOREIGN KEY(version_idx) REFERENCES file_versions(idx)
+) STRICT
+;
+CREATE TABLE IF NOT EXISTS file_names (
+    idx INTEGER NOT NULL PRIMARY KEY,
+    name TEXT NOT NULL UNIQUE COLLATE BINARY
+) STRICT
+;
+CREATE TABLE IF NOT EXISTS file_versions (
+    idx INTEGER NOT NULL PRIMARY KEY,
+    version TEXT NOT NULL UNIQUE COLLATE BINARY
 ) STRICT
 ;
 -- FileInfos store the actual protobuf object. We do this separately to keep
@@ -49,11 +61,17 @@ CREATE UNIQUE INDEX IF NOT EXISTS files_remote_sequence ON files (device_idx, re
     WHERE remote_sequence IS NOT NULL
 ;
 -- There can be only one file per folder, device, and name
-CREATE UNIQUE INDEX IF NOT EXISTS files_device_name ON files (device_idx, name)
-;
--- We want to be able to look up & iterate files based on just folder and name
-CREATE INDEX IF NOT EXISTS files_name_only ON files (name)
+CREATE UNIQUE INDEX IF NOT EXISTS files_device_name ON files (device_idx, name_idx)
 ;
 -- We want to be able to look up & iterate files based on blocks hash
 CREATE INDEX IF NOT EXISTS files_blocklist_hash_only ON files (blocklist_hash, device_idx) WHERE blocklist_hash IS NOT NULL
 ;
+-- We need to look by name_idx or version_idx for garbage collection.
+-- This will fail pre-migration for v4 schemas, which is fine.
+-- syncthing:ignore-failure
+CREATE INDEX IF NOT EXISTS files_name_idx_only ON files (name_idx)
+;
+-- This will fail pre-migration for v4 schemas, which is fine.
+-- syncthing:ignore-failure
+CREATE INDEX IF NOT EXISTS files_version_idx_only ON files (version_idx)
+;