CMSIS-NN: Update revision number & support info

1. Revision hisory is updated to reflect the interface change
2. Support link is added

Change-Id: I0fae3c9921b9d8cd646bb8b79daf371fc0318c95

ARM.CMSIS.pdsc

@@ -24,10 +24,11 @@
        - RTX4: Purged pre-built libs from Git
       CMSIS-RTOS2:
        - RTX5: Purged pre-built libs from Git
-      CMSIS-NN: 1.4.0 (see revision history for details)
+      CMSIS-NN: 2.0.0 (see revision history for details)
        - Major interface change for functions compatible with TensorFlow Lite for Microcontroller
        - Added optimization for SVDF kernel - DSP extension only
        - Improved MVE performance for fully Connected and max pool operator
+       - NULL bias support for fully connected operator in non-MVE case(Can affect performance)
        - Expanded unit test suite along with support for FVP
     </release>
     <release version="5.7.0" date="2020-04-09">

CMSIS/DoxyGen/NN/src/history.txt

@@ -7,7 +7,7 @@
     <th>Description</th>
   </tr>
     <tr>
-    <td>V1.4.0</td>
+    <td>V2.0.0</td>
     <td>
       Added the following function for int8 SVDF operator.<br>
       <ul>

CMSIS/NN/README.md

@@ -5,6 +5,9 @@ This page  give a quick overview of the functions available and key differences
 
 **Note:** The GitHub documentation does not follow the *develop* branch but rather the last official release in the *master* branch. Consequently, the group documentation linked to in the table table might not have the listed API. Please refer to the description in the [header](https://github.com/ARM-software/CMSIS_5/blob/develop/CMSIS/NN/Include/arm_nnfunctions.h) file instead.
 
+## Support / Contact
+For any questions or to reach the CMSIS-NN team, please create a new issue in https://github.com/ARM-software/CMSIS_5/issues
+
 ## Legacy vs TFL micro compliant APIs
 There are two kinds of APIs available in the CMSIS-NN repository; One that supports a legacy symmetric quantization scheme[1] and one that supports TFL micro's symmetric quantization scheme. One of the main differences is how the quantization is performed. The legacy APIs have a fixed point format with power of 2 scaling. This simplifies the re-quantization to a cycle efficient shift operation. No new development is done on the legacy functions and all of the new development is on the functions that support TFL micro. The table below highlights some of the differences between the two formats for convolution related functions. The TFL micro compliant APIs in most cases have a _s8 suffix and is always specified in the API header file.