- create ceilLog2() function based on floorLog2
- use intrinsics for MSVC in floorLog2
- replace all ceil(log2()) calls in HLS with ceilLog2()

- add option to completely disable
- add iparameter to software manual

  - Remove additional unused variable.

- In InterPrediction::rightShiftMSB, use existing floorLog2 function

- In addAvg_SSE, avoid unnecessary use of 32-bit path

- In copyBufferSimd, reduce number of loops

- In paddingSimd, take advantage of padding extent being either 1 or 2

- In addBIOAvg4_SSE, reduce number of operations and avoid nasty
Xmm register -> memory -> integer register path

Overall, a decoder runtime reduction of about 3% is expected

Note that lines 450-457 should already support the loosened
restriction

Use vectors and resize them if needed instead of deleting and
reallocating memory for every picture. This appears to have a
significant impact on decoder run time.

Use PLTRunMode type where appropriate instead of bool since
using vector<bool> creates a number of issues

This reduces the number of dynamic allocations

Allocate single large buffer instead of many small buffers for
holding scaling lists data

Remove partitioner factory to simplify memory allocation for the
partitioner (use stack instead of heap)

Overall decoder run time is reduced by about 5%