Trainer validates `gradient_clip_algorithm` although `configure_gradient_clipping` is defined

🐛 Bug

I can't pass a custom gradient clipping algorithm, although I implemented configure_gradient_clipping hook, Documentation and release notes hints that you can use configure_gradient_clipping to implement your custom gradient clipping algorithm (.release notes: ... This means you can now implement state-of-the-art clipping algorithms with Lightning! ...) .

Please, allow to pass custom algorithm names in gradient_clip_algorithm when the model implements configure_gradient_clipping.

To Reproduce

from pytorch_lightning import LightningModule, Trainer

class BoringModel(LightningModule):
    def __init__(self):
        super().__init__()
        self.layer = torch.nn.Linear(32, 2)

    def forward(self, x):
        return self.layer(x)

    def training_step(self, batch, batch_idx):
        loss = self(batch).sum()
        self.log("train_loss", loss)
        return {"loss": loss}

    def validation_step(self, batch, batch_idx):
        loss = self(batch).sum()
        self.log("valid_loss", loss)

    def test_step(self, batch, batch_idx):
        loss = self(batch).sum()
        self.log("test_loss", loss)

    def configure_optimizers(self):
        return torch.optim.SGD(self.layer.parameters(), lr=0.1)

    def configure_gradient_clipping(
        self,
        optimizer: Optimizer,
        optimizer_idx: int,
        gradient_clip_val: Optional[Union[int, float]] = None,
        gradient_clip_algorithm: Optional[str] = None,
    ):     
        if gradient_clip_algorithm == "my_custom_clipping_algorithm":
            my_custom_clipping(optimizer, gradient_clip_val, gradient_clip_algorithm)
        else:  # Lightning will handle the gradient clipping
            self.clip_gradients(
                optimizer,
                gradient_clip_val=gradient_clip_val,
                gradient_clip_algorithm=gradient_clip_algorithm
            )


trainer = Trainer(
        default_root_dir=os.getcwd(),
        limit_train_batches=1,
        limit_val_batches=1,
        limit_test_batches=1,
        num_sanity_val_steps=0,
        max_epochs=1,
        enable_model_summary=False,
        gradient_clip_algorithm="my_custom_clipping_algorithm"
    )

Expected behavior

Environment

• PyTorch Lightning Version: 1.5.4

Additional context

I want to train NFNets with Adaptive Gradient Clipping and compare with standard L2 gradient clipping.