v0.10.0 tracking issue

[coreylowman]

This will be a breaking issue to prepare for following releases.

• Add Weight Decay to optimizers #99
• Add tensor_ops::max_pool/avg_pool and nn::MaxPool/AvgPool #157
• Update examples to use Module::forward_mut #150
• Update Dropout and DropoutOnIn to use Module::forward_mut #149
• Adding nn::BatchNorm2D #228
• Add documentation for feature flags #229
• no_std support #201
• Move conv_forward and conv_backward to devices #198
• MultiHeadAttention needs to permute before reshape & after final matul #158
• Reorg map & binary map #147
• See if adding 2 muts and ref method in gradients can speed up binary map #146
• Add transpose/permute_axes operation #108
• Add reductions across multiple axes #163
• Add batched select #177
• Revamp prelude/imports #179
• Revamp examples #168
• Zero copy broadcasts #265
• Mergable tapes for use instead of moving aroudn the single tape #195
• Binary functions should take ownership of both operands #267