// Check for PDF - could be either key format
        if (queryMap.containsKey("labels.facet_filetype_pdf")) {
            assertEquals("filetype:pdf", queryMap.get("labels.facet_filetype_pdf"));
        } else {
            assertEquals("filetype:pdf", queryMap.get("PDF"));
        }

        // Check for XLS - could be either key format

This way, you don't have to know beforehand what the valid field/attribute names are (as would be the case with Pydantic models).

This would be useful if you want to receive keys that you don't already know.

---

Another useful case is when you want to have keys of another type (e.g., `int`).

That's what we are going to see here.

In this case, you would accept any `dict` as long as it has `int` keys with `float` values:

   *
   * The problem with these is that remove(K, long) has to be done in two phases by definition ---
   * first decrementing to zero, and then removing. putIfAbsent or replace could observe the
   * intermediate zero-state. Ways we could deal with this are:
   *
   * - Don't define any of the ConcurrentMap operations. This is the current state of affairs.
   *

    // This class can make no hard guarantees.  The methods in this class are inherently flaky, but
    // we try hard to make them robust in practice.  We could additionally try to add in a system
    // load timeout multiplier.  Or we could try to use a CPU time bound instead of wall clock time
    // bound.  But these ideas are harder to implement.  We do not try to detect or handle a
    // user-specified -XX:+DisableExplicitGC.

- Fix a bug that Pods could stuck in the unschedulable pod pool 
  if they're rejected by PreEnqueue plugins that could change its result by a change in resources apart from Pods.
  
  DRA plugin is the only plugin that meets the criteria of the bug in in-tree, 
  and hence if you have `DynamicResourceAllocation` feature flag enabled, 

     * calling us again in case of interrupt. Then, if we were to call await(...) immediately, as we
     * do in the other Uninterruptibles methods, it would throw immediately. Then we'd restore the
     * interrupt and return again, and the user would call us again, creating a busy wait.
     *
     * Thus, we need to clear the interrupt eagerly in case it's an interrupt from a previous call

      int result = BigIntegerMath.log2(x, HALF_UP);
      BigInteger x2 = x.pow(2);
      // x^2 < 2^(2 * result + 1), or else we would have rounded up
      assertThat(ZERO.setBit(2 * result + 1)).isGreaterThan(x2);
      // x^2 >= 2^(2 * result - 1), or else we would have rounded down
      assertTrue(result == 0 || ZERO.setBit(2 * result - 1).compareTo(x2) <= 0);
    }
  }

  public void testLog2HalfDown() {

     * to "pull in" transitive dependencies of eliminated projects, as for case above, the properly filtered list would
     * be {@code a -> c}.
     * <p>
     * Original code would falsely report {@code a} project as "without dependencies", basically would lose link due
     * filtering. This causes build ordering issues in concurrent builders.
     */
    private List<MavenProject> applyFilter(

      Stopwatch s = Stopwatch.createStarted();
      // here is where you would do something
      total += s.elapsed(NANOSECONDS);
    }
    return total;
  }

  @Benchmark
  long manual(int reps) {
    long total = 0;
    for (int i = 0; i < reps; i++) {
      long start = System.nanoTime();
      // here is where you would do something
      total += System.nanoTime() - start;
    }
    return total;
  }

   * its runtime check.
   *
   * <p>An example use case for this method is in implementing an {@code Iterator<T>} whose {@code
   * next} field is lazily initialized. The type of that field would be {@code @Nullable T}, and the
   * code would be responsible for populating a "real" {@code T} (which might still be the value
   * {@code null}!) before returning it to callers. Depending on how the code is structured, a