`IonReader.next()` throws `ArrayIndexOutOfBoundsException` for some corrupt content

[cowtowncoder]

(note: OSS-Fuzz sourced from https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=66131)

Looks like there's a code path that produces AIOOBE for corrupt content, with stack trace of:

 	== Java Exception: java.lang.ArrayIndexOutOfBoundsException: Index 16 out of bounds for length 16
	 at com.amazon.ion.impl.IonCursorBinary.uncheckedNextToken(IonCursorBinary.java:1571)
	 at com.amazon.ion.impl.IonCursorBinary.nextValue(IonCursorBinary.java:1714)
	 at com.amazon.ion.impl.IonReaderContinuableCoreBinary.nextValue(IonReaderContinuableCoreBinary.java:460)
	 at com.amazon.ion.impl.IonReaderContinuableApplicationBinary.nextValue(IonReaderContinuableApplicationBinary.java:942)
	 at com.amazon.ion.impl.IonReaderContinuableTopLevelBinary.handleIncompleteValue(IonReaderContinuableTopLevelBinary.java:150)
	 at com.amazon.ion.impl.IonReaderContinuableTopLevelBinary.next(IonReaderContinuableTopLevelBinary.java:166)
	 at com.fasterxml.jackson.dataformat.ion.IonParser.nextToken(IonParser.java:657)
	 at com.fasterxml.jackson.core.JsonParser.nextTextValue(JsonParser.java:1161)

[cowtowncoder]

Unable to reproduce at this point; most code paths / call sequences produce proper IonException.
Will try to tweak test case bit more to reproduce.

[cowtowncoder]

@tgregg This is for the 3rd OSS-Fuzz reported new-or-recycled issue, I was unable to reproduce.

@arthurscchan Maybe you can reproduce this issue? Idea is to reproduce but not add defensive programming yet, to let ion-java developers to handle at appropriate level

[tgregg]

#477 tweaks the test case to reproduce the problematic stack trace.

[cowtowncoder]

Test now fails properly. Thank you @tgregg !