Added documentation of the whole "Record and replay" feature to Doxygen docs and README.md.

src/vk_mem_alloc.h

@@ -67,6 +67,7 @@ Documentation of all members: vk_mem_alloc.h
     - [Memory initialization](@ref debugging_memory_usage_initialization)
     - [Margins](@ref debugging_memory_usage_margins)
     - [Corruption detection](@ref debugging_memory_usage_corruption_detection)
+  - \subpage record_and_replay
 - \subpage usage_patterns
   - [Simple patterns](@ref usage_patterns_simple)
   - [Advanced patterns](@ref usage_patterns_advanced)
@@ -920,6 +921,52 @@ Margin validation (corruption detection) works only for memory types that are
 `HOST_VISIBLE` and `HOST_COHERENT`.
 
 
+\page record_and_replay Record and replay
+
+\section record_and_replay_introduction Introduction
+
+While using the library, sequence of calls to its functions together with their
+parameters can be recorded to a file and later replayed using standalone player
+application. It can be useful to:
+
+- Test correctness - check if same sequence of calls will not cause crash or
+  failures on a target platform.
+- Gather statistics - see number of allocations, peak memory usage, number of
+  calls etc.
+- Benchmark performance - see how much time it takes to replay the whole
+  sequence.
+
+\section record_and_replay_usage Usage
+
+<b>To record sequence of calls to a file:</b> Fill in
+VmaAllocatorCreateInfo::pRecordSettings member while creating #VmaAllocator
+object. File is opened and written during whole lifetime of the allocator.
+
+<b>To replay file:</b> Use VmaReplay - standalone command-line program.
+Precompiled binary can be found in "bin" directory.
+Its source can be found in "src/VmaReplay" directory.
+Its project is generated by Premake.
+Command line syntax is printed when the program is launched without parameters.
+Basic usage:
+
+    VmaReplay.exe MyRecording.csv
+
+<b>Documentation of file format</b> can be found in file: "docs/Recording file format.md".
+It's a human-readable, text file in CSV format (Comma Separated Values).
+
+\section record_and_replay_additional_considerations Additional considerations
+
+- Replaying file that was recorded on a different GPU (with different parameters
+  like `bufferImageGranularity`, `nonCoherentAtomSize`, and especially different
+  set of memory heaps and types) may give different performance and memory usage
+  results, as well as issue some warnings and errors.
+- Current implementation of recording in VMA, as well as VmaReplay application, is
+  coded and tested only on Windows. Inclusion of recording code is driven by
+  `VMA_RECORDING_ENABLED` macro. Support for other platforms should be easy to
+  add. Contributions are welcomed.
+- Currently calls to vmaDefragment() function are not recorded.
+
+
 \page usage_patterns Recommended usage patterns
 
 See also slides from talk:
@@ -1198,7 +1245,7 @@ The library uses following algorithm for allocation, in order:
 Features deliberately excluded from the scope of this library:
 
 - Data transfer - issuing commands that transfer data between buffers or images, any usage of
-  `VkCommandList` or `VkCommandQueue` and related synchronization is responsibility of the user.
+  `VkCommandList` or `VkQueue` and related synchronization is responsibility of the user.
 - Allocations for imported/exported external memory. They tend to require
   explicit memory type index and dedicated allocation anyway, so they don't
   interact with main features of this library. Such special purpose allocations