StatusInfoReader cannot be re-opened after close() (SourceReader contract violation)

[thorwhalen]

Found while modernizing this repo's packaging (PR pending on claude/rollout-modernize).

Symptom

SourceReader's documented contract allows a reader to be closed and re-opened — its own class docstring demonstrates exactly that, with open_count reaching 2 on the second with block. StatusInfoReader cannot do this:

from pchealthstream2py.pchealth import StatusInfoReader
import time

r = StatusInfoReader(read_interval_ms=50)
with r:
    time.sleep(0.3)
with r:              # RuntimeError: threads can only be started once
    time.sleep(0.3)

Cause

StatusInfoReader is a threading.Thread (it subclasses both SourceReader and threading.Thread), and open() calls self.start(). A Thread object can only be started once, so the second open() always raises.

Suggested fix

Stop being a thread and have one instead: keep the polling loop in a plain function (or keep run() as the target) and have open() create a fresh threading.Thread(target=..., daemon=True) each time, storing it on the instance; close() sets the stop event and (optionally) joins it. That makes open/close/open work, keeps the buffering behaviour identical, and removes the remaining friction of inheriting Thread's API surface (start, join, is_alive, name, ...) into what is really a SourceReader.

This is a behaviour-visible change (isinstance(reader, threading.Thread) becomes False), which is why it was left out of the packaging-modernization PR rather than bundled into it.

Related, already fixed on claude/rollout-modernize

• the stop flag was named _stop, shadowing threading.Thread._stop and making join() raise TypeError: 'Event' object is not callable;
• _index / _data / _stop / _bt were class attributes, so all instances shared one queue and one stop flag.