[Snippets] PropagateSubtensors pass removed (#32316)

### Details:
`PropagateSubtensors` is a legacy error-prone (with hard workarounds and
high complexity) lowered pass used for Brgemm subtensors propagation in
static shapes scenario. Currently, this work is duplicated by Brgemm
executor whose code is much simpler. This means, that
`PropagateSubtensors` can be removed without any consequences
### Tickets:
 - *N\A*

Author: v-Golubev

src/common/snippets/include/snippets/lowered/pass/brgemm_blocking.hpp

@@ -32,8 +32,6 @@ namespace ov::snippets::lowered::pass {
 class BrgemmBlockingBase {
 public:
     virtual ~BrgemmBlockingBase() = default;
-    static snippets::lowered::SpecificIterationHandlers get_default_blocking_loop_handlers(size_t work_amount,
-                                                                                           size_t block_size);
 
 protected:
     /**

src/common/snippets/include/snippets/lowered/pass/propagate_subtensors.hpp

@@ -17,27 +17,13 @@
 #include "snippets/lowered/loop_info.hpp"
 #include "snippets/lowered/loop_manager.hpp"
 #include "snippets/lowered/loop_port.hpp"
-#include "snippets/lowered/pass/propagate_subtensors.hpp"
 #include "snippets/lowered/specific_loop_iter_handlers.hpp"
-#include "snippets/lowered/specific_loop_iter_types.hpp"
 #include "snippets/utils/utils.hpp"
 
 namespace ov::snippets::lowered::pass {
 using namespace ov::snippets::utils;
 using PortType = LoopPort::Type;
 
-lowered::SpecificIterationHandlers BrgemmBlockingBase::get_default_blocking_loop_handlers(size_t work_amount,
-                                                                                          size_t block_size) {
-    OPENVINO_ASSERT(block_size != 0, "block size must be non zero");
-    SpecificIterationHandlers handlers;
-    const auto tail_size =
-        utils::is_dynamic_value(work_amount) ? utils::get_dynamic_value<size_t>() : work_amount % block_size;
-    if (tail_size != 0) {
-        handlers.register_pass<lowered::SpecificLoopIterType::LAST_ITER, lowered::pass::UpdateSubtensors>(tail_size);
-    }
-    return handlers;
-}
-
 bool BrgemmBlockingBase::blocking_loop_exists(const lowered::LoopManagerPtr& loop_manager,
                                               const ExpressionPtr& brgemm_expr) {
     auto check_port = [&](const LoopPort& p) {
@@ -87,14 +73,17 @@ void BrgemmBlockingBase::mark_k_blocking(const LoopManagerPtr& loop_manager,
     loop_manager->get_loop_info<UnifiedLoopInfo>(id)->set_handlers(get_k_loop_handlers(k, block_size_k));
 }
 
-SpecificIterationHandlers BrgemmBlockingBase::get_m_loop_handlers(size_t work_amount, size_t block_size) const {
-    return get_default_blocking_loop_handlers(work_amount, block_size);
+SpecificIterationHandlers BrgemmBlockingBase::get_m_loop_handlers([[maybe_unused]] size_t work_amount,
+                                                                  [[maybe_unused]] size_t block_size) const {
+    return {};
 }
-SpecificIterationHandlers BrgemmBlockingBase::get_n_loop_handlers(size_t work_amount, size_t block_size) const {
-    return get_default_blocking_loop_handlers(work_amount, block_size);
+SpecificIterationHandlers BrgemmBlockingBase::get_n_loop_handlers([[maybe_unused]] size_t work_amount,
+                                                                  [[maybe_unused]] size_t block_size) const {
+    return {};
 }
-SpecificIterationHandlers BrgemmBlockingBase::get_k_loop_handlers(size_t work_amount, size_t block_size) const {
-    return get_default_blocking_loop_handlers(work_amount, block_size);
+SpecificIterationHandlers BrgemmBlockingBase::get_k_loop_handlers([[maybe_unused]] size_t work_amount,
+                                                                  [[maybe_unused]] size_t block_size) const {
+    return {};
 }
 
 std::tuple<size_t, size_t, size_t> BrgemmBlockingBase::get_blocking_params(const ExpressionPtr& brgemm_expr) const {

src/common/snippets/src/lowered/pass/brgemm_blocking.cpp

@@ -8,7 +8,6 @@
 #include <utility>
 
 #include "snippets/lowered/pass/iter_handler.hpp"
-#include "snippets/lowered/pass/propagate_subtensors.hpp"
 #include "snippets/utils/utils.hpp"
 
 namespace ov::snippets::lowered {
@@ -28,7 +27,6 @@ SpecificIterationHandlers::SpecificIterationHandlers(size_t loop_work_amount,
         }
         if (last_iter_increment != 0) {
             m_last_iter_handlers.register_pass<lowered::pass::UpdateMemoryAccessCounts>(last_iter_increment);
-            m_last_iter_handlers.register_pass<lowered::pass::UpdateSubtensors>(last_iter_increment);
             // Last Iterations of Loop processed last dimensions with Eltwise nodes inside should have increment = 1
             if (last_iter_increment == 1) {
                 m_last_iter_handlers.register_pass<lowered::pass::SetLoopIncrementOne>();

src/common/snippets/src/lowered/pass/propagate_subtensors.cpp

@@ -36,8 +36,7 @@ void create_gemm_loop_infos(const LinearIRPtr& linear_ir,
                 std::vector<LoopPort>{LoopPort::create<PortType::NotProcessed>(gemm_expr->get_input_port(0)),
                                       LoopPort::create<PortType::Incremented>(gemm_expr->get_input_port(1), 0)},
                 std::vector<LoopPort>{LoopPort::create<PortType::Incremented>(gemm_expr->get_output_port(0), 0)},
-                false,
-                BrgemmBlockingBase::get_default_blocking_loop_handlers(n, n_blk)));
+                false));
     }
     if (m_block) {
         std::vector<LoopPort> entries{LoopPort::create<PortType::Incremented>(gemm_expr->get_input_port(0), 1),
@@ -46,8 +45,7 @@ void create_gemm_loop_infos(const LinearIRPtr& linear_ir,
             std::make_shared<ov::snippets::lowered::UnifiedLoopInfo>(m, m_blk,
                 entries,
                 std::vector<LoopPort>{LoopPort::create<PortType::Incremented>(gemm_expr->get_output_port(0), 1)},
-                false,
-                BrgemmBlockingBase::get_default_blocking_loop_handlers(m, m_blk)));
+                false));
     }
 }