src: Current code produces warnings for possible value overflows. (#6665)

As the input values are uint8_t types, any shift may result in value loss.
Explicit promotion to the output type (uint32_t) keeps things safe.
Have also changed the int32_t in ads1220_read_adc to uint32_t, type
promotion and bit manipulation are a bit 'weird' on signed integers, so
keep it as an unsigned to align with following function call parameter type.
Have retained the prior explicit sign extension logic however.

Signed-off-by: Bevan Weiss <bevan.weiss@gmail.com>

src/sensor_ads1220.c

@@ -86,7 +86,9 @@ ads1220_read_adc(struct ads1220_adc *ads1220, uint8_t oid)
     barrier();
 
     // create 24 bit int from bytes
-    int32_t counts = (msg[0] << 16) | (msg[1] << 8) | msg[2];
+    uint32_t counts = ((uint32_t)msg[0] << 16)
+                    | ((uint32_t)msg[1] << 8)
+                    | ((uint32_t)msg[2]);
 
     // extend 2's complement 24 bits to 32bits
     if (counts & 0x800000)