Gujarati, Oriya ss2.1: clarify positioning of non-canonical decomposi…

…tions for top-and-right multipart matras. Also unify minor differences in 2.1 wording between the two.

opentype-shaping-gujarati.md

@@ -732,15 +732,20 @@ into their components. Gujarati has one multi-part dependent vowel,
 
 > "Candra O" (`U+0AC9`) decomposes to "`U+0AC5`,`U+0ABE`"
 
-> Note: "Candra O" is atypical in that it would decompose into one
-> above-base mark and one right-side mark. 
-
-> This combination, unlike other multi-part matras, does not affect
-> reordering: the two decomposed components would be tagged for the
-> `POS_AFTER_SUBJOINED` and `POS_AFTER_POST` sorting positions,
-> respectively, and neither will need to be reordered. In addition,
-> the decomposition is not canonical in Unicode. Consequently, shaping
-> engines may choose to skip it.
+> Note: "Candra O" is categorized in Unicode as being a top-and-right
+> matra, a combination that would normally decompose into one
+> TOP_POSITION mark and one RIGHT_POSITION mark. In "Candra O",
+> however, the `U+0AC5` component is intended to be positioned over the
+> `U+0ABE` component, not above the base.
+>
+> Consequently, the two decomposed components should both be tagged
+> for the `POS_AFTER_POST` sorting position, and neither will need to
+> be reordered.
+>
+> In addition, the decomposition is not canonical in
+> Unicode. so performing the decomposition may trigger unknown
+> behavior from other components of the software stack. Consequently,
+> shaping engines may choose to skip it.
 
 Because this decomposition is a character-level operation, the shaping
 engine may choose to perform it earlier, such as during an initial

opentype-shaping-oriya.md

@@ -803,22 +803,27 @@ has a canonical decomposition, so this step is unambiguous.
 >
 > "Au" (`U+0B4C`) decomposes to "`U+0B47`,`U+0B57`"
 
-Because this decomposition is a character-level operation, the shaping
-engine may choose to perform it earlier, such as during an initial
-Unicode-normalization stage. However, all such decompositions must be
-completed before the shaping engine begins step three, below.
-
 > Note: "Au Length Mark" (`U+0B57`) is categorized in Unicode as being a
 > top-and-right matra, a combination that would normally decompose
-> into one above-base mark and one right-side mark
-> (`U+0B3E`,`U+0B56`). This combination, unlike other multi-part
-> matras, does not affect reordering. 
-> 
+> into one TOP_POSITION mark and one RIGHT_POSITION mark
+> (`U+0B3E`,`U+0B56`). In "Au Length Mark", however, the `U+0B3E`
+> component is intended to be positioned over the `U+0B56` component,
+> not above the base.
+>
+> Consequently, the two decomposed components should both be tagged
+> for the `POS_AFTER_POST` sorting position, and neither will need to
+> be reordered.
+>
 > In addition, the decomposition is not canonical in
 > Unicode, so performing the decomposition may trigger unknown
 > behavior from other components of the software stack. Consequently,
 > shaping engines may choose to skip it. 
 
+Because this decomposition is a character-level operation, the shaping
+engine may choose to perform it earlier, such as during an initial
+Unicode-normalization stage. However, all such decompositions must be
+completed before the shaping engine begins step three, below.
+
 ![Two-part matra decomposition](/images/oriya/oriya-matra-decompose.png)
 
 #### 2.3: Tag matras ####