JDBC backend: isExistsTable lists every table of the database, once per tree, on every backend open

[vharseko]

Describe the bug

isExistsTable() asks the catalog for every table of the database and looks for a match in Java — no catalog filter, no schema filter, no table name:

// opendj-server-legacy/src/main/java/org/opends/server/backends/jdbc/JDBCStorage.java:269
try (final ResultSet rs = con.getMetaData().getTables(null, null, null, new String[]{"TABLE"})) {
    while (rs.next()) {
        if (tree2table.get(treeName).equalsIgnoreCase(rs.getString("TABLE_NAME"))) {

It is called from openTree(createOnDemand) — through AbstractTree.open(), which every tree of the backend goes through — and from deleteTree(). So opening a backend walks the whole catalog once per tree: about 25 times for a stock suffix, and the cost is O(trees × tables in the database) for what ought to be a single indexed lookup. A JDBC backend sharing its database with another application pays for that application's tables too, on every open.

The same two DatabaseMetaData calls of this class — getTables() here and getIndexInfo() in isExistsIndex() — also take no setQueryTimeout, so they are outside the statement bounds of #877/#882 and can only be bounded by the read timeout asked for in #885. Making this one cheap is the fix that matters; the bound is the smaller half.

Impact

The open of a backend, and dsconfig create-backend-index on a running server, are slower than they need to be by a factor that grows with what else lives in the database — a cost that is invisible in a test database and grows quietly in a shared production one.

Expected behavior

• the table name is passed to getTables() as the pattern, so the driver asks its catalog for one row rather than all of them;
• the name is passed in the form the catalog stores it — an unquoted identifier is folded to upper case on Oracle and to lower case on PostgreSQL, and a metadata pattern is matched against the stored form — asked of the driver through storesUpperCaseIdentifiers() / storesLowerCaseIdentifiers() rather than by matching its class name, as the dialect switches elsewhere in this class do;
• the name of the row that comes back is still compared, since _ is a single-character wildcard in a metadata pattern.

Environment

master (5.2.x), all four JDBC dialects (PostgreSQL, MySQL, Oracle, MS SQL Server).

Split out of #885, which keeps the two bounds that live in CachedConnection.