Had this branch hanging around and after chat with Jakub we
figured its worth having for when we need to figure out
client providers

Bumps [ollama](https://github.com/ollama/ollama-python) from 0.4.5 to 0.4.6.
- [Release notes](https://github.com/ollama/ollama-python/releases)
- [Commits](ollama/ollama-python@v0.4.5...v0.4.6)

---
updated-dependencies:
- dependency-name: ollama
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>

Bumps [litellm](https://github.com/BerriAI/litellm) from 1.57.10 to 1.58.0.
- [Release notes](https://github.com/BerriAI/litellm/releases)
- [Commits](BerriAI/litellm@v1.57.10...v1.58.0)

---
updated-dependencies:
- dependency-name: litellm
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>

The build-arg prevents that the step for downloading the UI code is cached.
This way we can ensure that we are always using the latest releasee from
the UI when building the image locally. The build-arg is already present
in the GH action to build and publish the image.

Additionally, remove the `secret` provided to the GH action to build the
image. It was used when the containers were not public

Add build-arg in `make image-build` command

Log User-agents from clients

Bump ollama from 0.4.5 to 0.4.6

Bump litellm from 1.57.10 to 1.58.0

Closes: #519

When using `/explain` function in copilot we were double counting th secrets.
The problem was that we were getting several `user` messages after the last
`assistant` message. We are using the last `assistant` message as means to identify
the user messages. Here is an example of how the request looked like for `/explain`:
```
[
  {"role": "assistant", "content": "some content"},
  {"role": "user", "content": content_with_secrets},
  {"role": "user", "content": content_with_secrets},
]
```

To avoid double counting now we check which was the secrets that matched after
the last `assistant` message and only consider the unique ones.

fix: Do not double count secrets on `/explain` copilot function

The nginx configuration uses a lot of defaults that work well in
server-like environments. However, this is not optimized for a local
dev-environment or single-container as codegate is meant to run.

The nginx server spawns several processes which mostly remain idle. This
is wasteful and it is taking over resources that could be used for
something else. The intention of this new configuration is to minimize
footprint.

Signed-off-by: Juan Antonio Osorio <ozz@stacklok.com>

There is no use in keeping this variable as it's only used for the
container tag. Releases use another tagging mechanism.

Signed-off-by: Juan Antonio Osorio <ozz@stacklok.com>

This leverages the package metadata instead of overwriting it with a
hardcoded version and description.

Signed-off-by: Juan Antonio Osorio <ozz@stacklok.com>

Add all-in-one minimal nginx config

Related: #583

We had been using a single DB schema that didn't change until now.
This introduces migrations using `alembic`. To create a new migration
one can use:
```sh
alembic revision -m "My migration"
```
That should generate an empty migration file that needs to be hand-filled.
Specifically the `upgrade` method which will be the one executed when
running the migration.
```python
"""My migration

Revision ID: <some_hash>
Revises: <previous_hash>
Create Date: YYYY-MM-DD HH:MM:SS.XXXXXX
"""
from alembic import op
import sqlalchemy as sa

revision = '<some_hash>'
down_revision = '<previous_hash>'
branch_labels = None
depends_on = None

def upgrade():
    pass

def downgrade():
    pass
```

Get version dynamically as opposed to the hardcoded one we have now

Use `latest` tag to build local codegate container

feat: Introduce DB migrations