conversions: Make NEXT_ID a true singleton.

We now instantiate NEXT_ID in sequencer.py, which avoids
having multiple modules make multiple copies of a sequencer
and possibly causing id collisions.

zerver/data_import/hipchat.py

@@ -40,10 +40,7 @@ from zerver.data_import.import_util import (
 
 from zerver.data_import.hipchat_attachment import AttachmentHandler
 from zerver.data_import.hipchat_user import UserHandler
-from zerver.data_import.sequencer import sequencer
+from zerver.data_import.sequencer import NEXT_ID
-
-# Create one sequencer for our entire conversion.
-NEXT_ID = sequencer()
 
 # stubs
 ZerverFieldsT = Dict[str, Any]

zerver/data_import/hipchat_attachment.py

@@ -9,10 +9,7 @@ from zerver.data_import.import_util import (
 
 from typing import Any, Dict, List, Optional
 
-from zerver.data_import.sequencer import sequencer
+from zerver.data_import.sequencer import NEXT_ID
-
-# Create one sequencer for our entire conversion.
-NEXT_ID = sequencer()
 
 class AttachmentHandler:
     def __init__(self) -> None:

zerver/data_import/sequencer.py

@@ -26,7 +26,7 @@ def sequencer() -> Callable[[str], int]:
         Use like this:
 
         NEXT_ID = sequencer()
-        message_id = NEXT_ID('message_id')
+        message_id = NEXT_ID('message')
     '''
     seq_dict = dict()  # type: Dict[str, Callable[[], int]]
 
@@ -37,3 +37,14 @@ def sequencer() -> Callable[[str], int]:
         return seq()
 
     return next_one
+
+'''
+NEXT_ID is a singleton used by an entire process, which is
+almost always reasonable.  If you want to have two parallel
+sequences, just use different `name` values.
+
+This object gets created once and only once during the first
+import of the file.
+'''
+
+NEXT_ID = sequencer()