Fix ByteContainer.setBytes to reallocate data when the incoming length
differs from the currently stored length.

Upstream licel/jcardsim master drops `this.length != length` from the
reallocation guard; the DT fork kept it. Without the check, calling
setKey(buf13, 0, 13) then setKey(buf64, 32, 32) on the same key reuses the
already-sized-13 internal data[] and triggers ArrayIndexOutOfBoundsException
in the subsequent Util.arrayCopy.

This regression hits HKDF-Extract-then-Expand on the same HMACKey, which is
exactly the shape Aliro §8.3.1.4 / §8.3.1.13 require.

diff --git a/src/main/java/com/licel/jcardsim/crypto/ByteContainer.java b/src/main/java/com/licel/jcardsim/crypto/ByteContainer.java
--- a/src/main/java/com/licel/jcardsim/crypto/ByteContainer.java
+++ b/src/main/java/com/licel/jcardsim/crypto/ByteContainer.java
@@ -106,7 +106,7 @@ public final class ByteContainer {
      * @param length length of data in byte array
      */
     public void setBytes(byte[] buff, short offset, short length) {
-        if (data == null) {
+        if (data == null || this.length != length) {
             switch (memoryType) {
                 case JCSystem.MEMORY_TYPE_TRANSIENT_DESELECT:
                     data = JCSystem.makeTransientByteArray(length, JCSystem.CLEAR_ON_DESELECT);