UHEPRNG - Ultra High Entropy Pseudo-Random Number Generator
                           Gibson Research Corporation
   ============================================================================
   LICENSE AND COPYRIGHT:  THIS CODE IS HEREBY RELEASED INTO THE PUBLIC DOMAIN
   Gibson Research Corporation releases and disclaims ALL RIGHTS AND TITLE IN
   THIS CODE OR ANY DERIVATIVES. Anyone may be freely use it for any purpose.
   ============================================================================
   This is GRC's cryptographically strong PRNG (pseudo-random number generator)
   for JavaScript. It is driven by 1536 bits of entropy, stored in an array of
   48, 32-bit JavaScript variables.  Since many applications of this generator,
   including ours with the "Off The Grid" Latin Square generator, may require
   the deteriministic re-generation of a sequence of PRNs, this PRNG's initial
   entropic state can be read and written as a static whole, and incrementally
   evolved by pouring new source entropy into the generator's internal state.
   ----------------------------------------------------------------------------
   ENDLESS THANKS are due Johannes Baagoe for his careful development of highly
   robust JavaScript implementations of JS PRNGs.  This work was based upon his
   JavaScript "Alea" PRNG which is based upon the extremely robust Multiply-
   With-Carry (MWC) PRNG invented by George Marsaglia. MWC Algorithm References:
   http://www.GRC.com/otg/Marsaglia_PRNGs.pdf
   http://www.GRC.com/otg/Marsaglia_MWC_Generators.pdf
   ----------------------------------------------------------------------------
   The quality of this algorithm's pseudo-random numbers have been verified by
   multiple independent researchers. It handily passes the fermilab.ch tests as
   well as the "diehard" and "dieharder" test suites.  For individuals wishing
   to further verify the quality of this algorithm's pseudo-random numbers, a
   256-megabyte file of this algorithm's output may be downloaded from GRC.com,
   and a Microsoft Windows scripting host (WSH) version of this algorithm may be
   downloaded and run from the Windows command prompt to generate unique files
   of any size:
   The Fermilab "ENT" tests: http://fourmilab.ch/random/
   The 256-megabyte sample PRN file at GRC: https://www.GRC.com/otg/uheprng.bin
   The Windows scripting host version: https://www.GRC.com/otg/wsh-uheprng.js
   ----------------------------------------------------------------------------
   Qualifying MWC multipliers are: 187884, 686118, 898134, 1104375, 1250205,
   1460910 and 1768863. (We use the largest one that's < 2^21)

   https://www.grc.com/otg/uheprng.htm