Creating groundwork for multiembedding models like insightface

[deven96]

related to #227

• Introduced the ModelResponse type which splits between models that return one output for one input and models that return multiple output for one input
• Upgraded Rust to 1.92.0
• Fixed SingleInputToEmbedding to reflect the fact that embeddings returned can be multiples

[github-actions]

Test Results

196 tests   196 ✅  9m 1s ⏱️
 31 suites    0 💤
  3 files      0 ❌

Results for commit 0dc3f81.

♻️ This comment has been updated with latest results.

[github-actions]

Benchmark Results

group                                                        main                                   pr
-----                                                        ----                                   --
store_batch_insertion_without_predicates/size_100            1.00   795.0±11.89µs        ? ?/sec    1.01    802.1±2.68µs        ? ?/sec
store_batch_insertion_without_predicates/size_1000           1.00      6.8±0.02ms        ? ?/sec    1.03      7.0±0.04ms        ? ?/sec
store_batch_insertion_without_predicates/size_10000          1.00     83.8±0.21ms        ? ?/sec    1.02     85.3±0.30ms        ? ?/sec
store_batch_insertion_without_predicates/size_100000         1.00    846.0±3.68ms        ? ?/sec    1.02    864.6±3.60ms        ? ?/sec
store_retrieval_no_condition/size_100                        1.02   1706.6±3.19µs        ? ?/sec    1.00   1674.6±4.12µs        ? ?/sec
store_retrieval_no_condition/size_1000                       1.01     11.8±0.02ms        ? ?/sec    1.00     11.7±0.07ms        ? ?/sec
store_retrieval_no_condition/size_10000                      1.02    117.2±0.15ms        ? ?/sec    1.00    115.0±0.23ms        ? ?/sec
store_retrieval_no_condition/size_100000                     1.02   1177.9±0.75ms        ? ?/sec    1.00   1158.1±5.18ms        ? ?/sec
store_retrieval_non_linear_kdtree/size_100                   1.02   1802.6±3.83µs        ? ?/sec    1.00   1767.2±4.51µs        ? ?/sec
store_retrieval_non_linear_kdtree/size_1000                  1.00     13.0±0.05ms        ? ?/sec    1.00     13.0±0.07ms        ? ?/sec
store_retrieval_non_linear_kdtree/size_10000                 1.01    129.9±0.16ms        ? ?/sec    1.00    128.7±0.90ms        ? ?/sec
store_retrieval_non_linear_kdtree/size_100000                1.01   1313.4±2.86ms        ? ?/sec    1.00   1298.4±7.58ms        ? ?/sec
store_sequential_insertion_without_predicates/size_100       1.02   1133.7±1.76µs        ? ?/sec    1.00   1111.4±2.89µs        ? ?/sec
store_sequential_insertion_without_predicates/size_1000      1.02     11.3±0.00ms        ? ?/sec    1.00     11.1±0.01ms        ? ?/sec
store_sequential_insertion_without_predicates/size_10000     1.02    113.0±0.05ms        ? ?/sec    1.00    111.0±0.06ms        ? ?/sec
store_sequential_insertion_without_predicates/size_100000    1.02   1129.9±0.61ms        ? ?/sec    1.00   1109.4±0.39ms        ? ?/sec

[Iamdavidonuh]

lGTM. Please add a TODO to document Single and Multiple Embeddings