Remove warning which only applied to Cohort Extractor

docs/data-sources/systmone.md

@@ -72,7 +72,10 @@ IMD rankings are recalculated every 3 or 4 years.
 Consequently, there is no original ranking for some newly built residences.
 In these cases, the rounded ranking is -1.
 
----8<-- 'includes/imd-warning-header.md'
+!!! warning
+    The original IMD ranking is rounded to the nearest 100 in the OpenSAFELY-TPP and OpenSAFELY-EMIS databases.
+    The rounded IMD ranking ranges from 0 to 32,800.
+    If there is no original ranking, then the rounded ranking is -1 in the OpenSAFELY-TPP database and `NULL` in the OpenSAFELY-EMIS database.
 
 Occasionally, a patient has multiple active registrations on a given date.
 If so, the address is chosen as follows:

docs/legacy/study-def-tricks.md

@@ -362,7 +362,16 @@ In addition:
 
 ## Grouping IMD by quintile
 
----8<-- 'includes/imd-warning-header.md'
+!!! warning
+    The original IMD ranking is rounded to the nearest 100 in the OpenSAFELY-TPP and OpenSAFELY-EMIS databases.
+    The rounded IMD ranking ranges from 0 to 32,800.
+    If there is no original ranking, then the rounded ranking is -1 in the OpenSAFELY-TPP database and `NULL` in the OpenSAFELY-EMIS database.
+
+!!! warning
+    Avoid extracting the rounded IMD ranking to a binary format, such as `.feather` or `.dta`.
+    Either nest it within a variable,
+    such as when [grouping rounded IMD by quintile](https://docs.opensafely.org/legacy/study-def-tricks/#grouping-imd-by-quintile),
+    or extract it to a non-binary format, such as `.csv.gz`.
 
 To group IMD by quintile, then: