Use scaled integer array (ifld) in OpenJPEG version of enc_jpeg2000

[DusanJovic-NOAA]

This PR fixes a bug in OpenJPEG version of enc_jpeg2000 routine by using scaled integer array instead of packed character buffer.

Fixes #518

[DusanJovic-NOAA]

Test tst_jpeg tests these functions. This test currently pass because the range of values being encoded is such that scaled integers fit in one byte. However, if I make these changes in tst_jpeg:

diff --git a/tests/tst_jpeg.c b/tests/tst_jpeg.c
index 40789ce..ca24378 100644
--- a/tests/tst_jpeg.c
+++ b/tests/tst_jpeg.c
@@ -80,7 +80,7 @@ main()

         printf("Testing jpcpack()/jpcunpack() call...");
         {
-            float fld[DATA_LEN] = {1.0, 2.0, 3.0, 0.0};
+            float fld[DATA_LEN] = {100.0, 200.0, 300.0, 0.0};
             float fld_in[DATA_LEN];
             g2int lcpack = PACKED_LEN;
             g2int idrstmpl[7] = {0, 1, 1, 16, 0, 0, 0};
@@ -103,7 +103,7 @@ main()
         printf("ok!\n");
         printf("Testing g2c_jpcpackd()/g2c_jpcunpackd() call...");
         {
-            double fld[DATA_LEN] = {1.0, 2.0, 3.0, 0.0};
+            double fld[DATA_LEN] = {10000.0, 20000.0, 30000.0, 0.0};
             double fld_in[DATA_LEN];
             size_t lcpack_st = PACKED_LEN;
             int idrstmpl[7] = {0, 1, 1, 16, 0, 0, 0};
@@ -125,7 +125,7 @@ main()
         printf("ok!\n");
         printf("Testing g2c_jpcpackf()/g2c_jpcunpackf() call...");
         {
-            float fld[DATA_LEN] = {1.0, 2.0, 3.0, 0.0};
+            float fld[DATA_LEN] = {1000.0, 2000.0, 3000.0, 0.0};
             float fld_in[DATA_LEN];
             size_t lcpack_st = PACKED_LEN;
             int idrstmpl[7] = {0, 1, 1, 16, 0, 0, 0};

test fails in current develop branch.

I'll update this test in my branch.