[Strategy backtesting] Add portfolio value tracking of agents + Streamlit

[gabrielfior]

When selecting a strategy, we observe the profitability of each (strategy, agent) combination, as well as other financial metrics (https://github.com/gnosis/prediction-market-agent-tooling/pull/548/files). However, one missing ingredient is the value that the portfolio has across time.

For example, if I placed a $10 bet on YES on market A on day 1 and now it's day 10 (and my portfolio consists only of this one bet), I can track the value of my portfolio between days 1-10. This allows me to follow strategies that depend on the current value of each user_position.
A visualization that goes in this direction (but not completely equivalent) can be seen in the screenshot below (max/min/realizable balance). While we can have a good idea of the current value of the portfolio by analyzing min/max vs time, it would be nice to have a concrete number on each block so that we could take more informed decisions, portfolio-wise.

Deliverables
-> Expand the match_bets_with_langfuse_traces to also export a portfolio_df tracking user_position balances vs time
-> (Optional) Create a Streamlit app that plots the historical portfolio per agent

[kongzii]

sounds good!

-> (Optional) Create a Streamlit app that plots the historical portfolio per agen

This could be added to the existing streamlit monitor app: https://github.com/gnosis/prediction-market-agent-tooling/blob/main/examples/monitor/monitor.py