We want to allow scale-up of memory when the checkpoint is executed on a machine with less memory than the expected available memory on the production machine, after restore.

Following the conservative approach in (abandoned) JDK-8204088, rather than trying to address all usages of `MaxHeapSize` VM option we might impose an additional limit — `CRaCMaxHeapSizeBeforeCheckpoint` — that will be applied until checkpoint. During restore this limitation will be removed to continue execution with the original `MaxHeapSize`. Since the default `MaxHeapSize` is set based on the perceived machine available memory we expect that both VM options would be set when the process is started.

For simplicity it is not expected that the value of `CRaCMaxHeapSizeBeforeCheckpoint` will be presented in existing memory APIs (e.g. `MemoryMxBean`); we will simply prevent heap from growing past this threshold.