Exercise 1: A Real, Catchable Page Fault — Possible Solution
====================================================================

THE TEST
------------------------------
    mem = PhysicalMemoryFramed(num_frames=4)
    pt = PageTable(page_size=64)
    pfn = mem.alloc_frame()
    pt.map_page(vpn=0, pfn=pfn)

    try:
        pt.translate(64 * 3)   # virtual page 3 -- never mapped
        fault_raised = False
    except PageFault as e:
        fault_raised = True
        print(e)

    valid = pt.translate(5)   # a genuinely mapped address, right next door

RESULT
------------------------------
    translating an unmapped virtual address raised: no mapping for
    virtual page 3 (virtual address 192)
    translating a genuinely mapped address still works fine: 5

fault_raised is True, and the second translate() call (a real, mapped
address) succeeds normally and returns the correct physical address.

WHY THE FAULT IS RAISED CLEANLY, NOT AS A KEYERROR
------------------------------
translate()'s own logic checks membership explicitly before ever
touching self.table[vpn]:

    if vpn not in self.table:
        raise PageFault(f"no mapping for virtual page {vpn}")
    pfn = self.table[vpn]

Without that explicit check, self.table[vpn] on an unmapped vpn would
raise a raw Python KeyError instead -- technically also an exception,
but one that exposes translate()'s own internal dict-based
implementation detail to anyone catching it, rather than a real,
purpose-built PageFault type that means something specific: "this
virtual address has no mapping." A caller catching PageFault
specifically doesn't need to know or care that a dict is involved at
all.

WHY THE NEIGHBORING VALID ADDRESS STILL WORKS
------------------------------
Virtual address 5 belongs to virtual page 0 (5 // 64 == 0), which IS
mapped -- to pfn, the frame allocated at the start of the test. This
confirms the PageFault for address 192 is specific to that one
unmapped page, not a sign that the whole page table or the
translate() method itself is broken. A single missing mapping doesn't
poison translation for every other, genuinely valid address.

WHY THIS WORKS AS AN ANSWER
------------------------------
Testing both the failing case (a real page fault, correctly typed and
catchable) and a neighboring success case in the same test confirms
the fault-handling code path is deliberate and precise -- it fires
exactly when it should, and only then, rather than being a fragile
side effect of how the page table happens to be implemented
internally.