fix(charts): leave the suspect and active runs out of the bar charts

The second review of PR #12 (after the suspect-run fix) found that the
two BAR charts were still plotting runs they should not have been, while
the curve chart and the tiles had been corrected.  The page was
contradicting itself.

The ABV chart plotted the suspect run
--------------------------------------
Run 12's impossible reading gives it an ABV of 18.64%.  The chart's y
axis went to 18.64, so the fifteen real bars were squashed into the
bottom ~45% of the plot - while the "Strongest 8.53%" tile on the same
page showed the correct figure.  The chart now leaves the suspect runs
out.

The attenuation chart plotted the active runs
----------------------------------------------
An active run's attenuation is measured against its CURRENT gravity, so
a brew that has just started shows a 0% bar.  That is the absence of a
result, not a result, and it sat directly above the table of finished
brews.  The chart now leaves the active runs out.

Verified against the live data:

    ABV chart bars:        bars=15 max=8.53   (was 16 bars, max 18.64)
    attenuation chart bars: bars=15 min=76.2   (was 17 bars, min 0)

A suspect active run no longer vanishes
---------------------------------------
The API flags suspect runs over EVERY run, so the charts leave them out.
But the analysis page only looked at the finished ones, so an active run
with a bad reading would be dropped from the charts and never listed in
"Readings that need a second look" - it would disappear with no
explanation anywhere.  The analysis page now lists suspect active runs
too.

The figures say what they are based on
--------------------------------------
- The analysis prose counted the non-suspect runs, not the ones that
  actually contributed a figure to the mean.  It now counts the latter.
- The history page's "Average ABV — across finished brews" was averaging
  15 of 16 without saying so, while the analysis page said "16 brews (15
  counted)".  It now says "across 15 of 16 finished brews".
- `/api/runs/<run_id>` carries the `suspect` and `active` flags that
  `/api/runs` does.

148 tests (up from 142).

src/flask/views/brews.py

@@ -198,6 +198,10 @@ def api_runs():
     # The runs whose readings are physically impossible.  They are flagged
     # rather than dropped: the charts use the flag to leave them out of the
     # averages and the curve envelope, but a caller can still see them.
+    #
+    # The flag is computed over EVERY run, not just the finished ones, so
+    # it agrees with what the charts do.  An active run with a bad reading
+    # is still a run with a bad reading.
     suspect_ids = {s["beer_run_id"] for s in implausible_runs(runs)}
 
     payload = []
@@ -228,6 +232,7 @@ def api_run(run_id: str):
     # The same flag the list endpoint carries, so a consumer of the detail
     # endpoint can tell an impossible run from a good one.
     summary["suspect"] = bool(implausible_runs([run]))
+    summary["active"] = run.is_active()
     return jsonify(summary)