Add double casts to silence MSVC C4244 warnings

aom\av1\encoder\pickrst.c(1195): warning C4244: 'initializing':
conversion from 'int64_t' to 'const double', possible loss of data

The double type has a 53-bit significand precision, so a loss of
precision is expected when a large int64_t value is converted to double.

Bug: oss-fuzz:566951534
Change-Id: I863901c448f0b1aec64dfc41f1078e64116ebafd

av1/encoder/pickrst.c

diff --git a/av1/encoder/pickrst.c b/av1/encoder/pickrst.c
index bc45960..304644d 100644
--- a/av1/encoder/pickrst.c
+++ b/av1/encoder/pickrst.c
@@ -1192,8 +1192,8 @@
       // Use floating-point arithmetic.
       for (int i = k; i < n - 1; i++) {
         if (A[k * stride + k] == 0) return 0;
-        const double c = A[(i + 1) * stride + k];
-        const double cd = A[k * stride + k];
+        const double c = (double)A[(i + 1) * stride + k];
+        const double cd = (double)A[k * stride + k];
         for (int j = 0; j < n; j++) {
           A[(i + 1) * stride + j] -= (int64_t)(A[k * stride + j] * c / cd);
         }