Exercise 1: Re-Running the Benchmark at 500 Calls — Possible Solution
====================================================================

WHAT CHANGED
------------------------------
Only the loop count N was changed, from 200 to 500 - the direct call
function, the HTTP server, and the timing method (time.perf_counter())
are all identical to this chapter's own benchmark.

RESULTS AT 500 CALLS
------------------------------
    500 direct calls: 0.04 ms total, 0.08 microseconds/call
    500 HTTP calls: 427.24 ms total, 854.48 microseconds/call
    ratio: 10762x slower

COMPARING TO THIS CHAPTER'S OWN 200-CALL RESULT
------------------------------
    200-call result: 0.11 microseconds/call direct, 1253.57 microseconds/call HTTP, ~11,661x
    500-call result: 0.08 microseconds/call direct, 854.48 microseconds/call HTTP, ~10,762x

Both per-call timings stayed in the same rough order of magnitude as
this chapter's own result - direct calls stayed well under 1
microsecond, and HTTP calls stayed in the roughly 800-1,300
microsecond range. The exact ratio moved from ~11,661x to ~10,762x,
which is expected: per-call network timings on a real machine vary run
to run (background load, OS scheduling, connection setup overhead), so
the precise multiplier isn't a fixed constant - but the ORDER OF
MAGNITUDE (thousands of times slower) held steady across both sample
sizes.

WHY THIS CONFIRMS THE FINDING WASN'T A FLUKE
------------------------------
A single 200-call run could plausibly have been an outlier. Re-running
at more than double the sample size and landing in the same
thousands-of-times-slower range confirms this chapter's own headline
finding is a genuine, repeatable property of crossing a network
boundary - not a one-off measurement artifact.

WHY THIS WORKS AS AN ANSWER
------------------------------
The benchmark is re-run with only the sample size changed, keeping
every other variable identical to this chapter's own setup, and the
new result is compared directly against the original rather than
reported in isolation - showing the finding replicates at a different
scale rather than just restating the original number.